feat(brain): garden, mirrors, vault-sync, provider registry consolidation
- .aurelio/garden/: model + agent garden (Google Cloud entries) - .aurelio/mirrors/: sync-mirrors.yaml + state tracking - .aurelio/skills/gcp/: Google Cloud skill - Consolidation audit + execution plan (2026-07-30) - vault-sync.py: Obsidian → GBrain MCP ingestion daemon - brain-to-gbrain.py: brain → GBrain migration tool - Provider registry + dist mirrors updated - .gitignore: exclude .runner, .mimocode/.cron-lock, drift/target Co-authored-by: Álvaro de Campos <campos@portugalfuturista.org>
This commit is contained in:
Raphael Cautus (Maestro)
parent
c96b6631ce
commit
749432fefc
22 changed files with 3406 additions and 590 deletions
|
|
@ -51,16 +51,19 @@ connectors:
|
|||
notes: DEFAULT forge (self-hosted first policy). LAN origin 192.168.0.9:3001. CLI = tea. Tokens in Vaultwarden.
|
||||
|
||||
- id: github
|
||||
label: GitHub
|
||||
label: GitHub — upstream mirror source
|
||||
category: devops
|
||||
kind: api-key
|
||||
auth: api_key
|
||||
endpoint: https://api.github.com
|
||||
surfaces: [mcp, theia]
|
||||
status: declared
|
||||
scopes: []
|
||||
scopes: [repo, read:org]
|
||||
host: null
|
||||
notes: Only for public mirrors/upstreams — Forgejo is the default forge. CLI = gh.
|
||||
notes: |
|
||||
Upstream source for the mirror pipeline (GitHub → Forgejo) and public mirrors.
|
||||
Forgejo is the default forge. PAT (classic, repo+read:org scope) in GITHUB_MIRROR_TOKEN.
|
||||
Configured in .aurelio/mirrors/sync-mirrors.yaml under targets.forgejo. CLI = gh.
|
||||
|
||||
- id: kubernetes
|
||||
label: Kubernetes
|
||||
|
|
@ -141,11 +144,35 @@ connectors:
|
|||
kind: oauth
|
||||
auth: oauth2
|
||||
endpoint: https://cloudresourcemanager.googleapis.com
|
||||
surfaces: [mcp, theia]
|
||||
status: declared
|
||||
scopes: [cloud-platform]
|
||||
surfaces: [mcp, theia, portal]
|
||||
status: provisioned
|
||||
scopes:
|
||||
- cloud-platform
|
||||
- https://www.googleapis.com/auth/gmail.readonly
|
||||
- https://www.googleapis.com/auth/calendar.readonly
|
||||
- https://www.googleapis.com/auth/userinfo.profile
|
||||
- https://www.googleapis.com/auth/contacts.readonly
|
||||
- https://www.googleapis.com/auth/cloud-platform.read-only
|
||||
host: null
|
||||
notes: CLI = gcloud. No real self-hosted equivalent; firebase emulators / fake-gcs-server for dev.
|
||||
notes: |
|
||||
PF default Google Cloud project civil-rarity-497813-r6 (€25k credits).
|
||||
Vertex AI Model Garden + Gemini APIs enabled; OAuth client configured for Workspace research ingestion.
|
||||
CLI = gcloud. Service-account fallback via GOOGLE_APPLICATION_CREDENTIALS for batch/background jobs.
|
||||
Self-hosted dev: firebase emulators / fake-gcs-server only; no full GCP emulator.
|
||||
|
||||
- id: google-cloud-vertex
|
||||
label: Google Cloud Vertex AI (Model Garden)
|
||||
category: cloud
|
||||
kind: api-key
|
||||
auth: api_key
|
||||
endpoint: https://us-central1-aiplatform.googleapis.com
|
||||
surfaces: [mcp, theia, portal]
|
||||
status: provisioned
|
||||
scopes: []
|
||||
host: null
|
||||
notes: |
|
||||
Vertex AI enterprise inference endpoint. Uses GOOGLE_CLOUD_API_KEY or service-account.
|
||||
See .aurelio/garden/model-garden/google-cloud.yaml for model IDs and context windows.
|
||||
|
||||
- id: azure
|
||||
label: Microsoft Azure
|
||||
|
|
@ -234,16 +261,50 @@ connectors:
|
|||
notes: Entra ID app registration required; exposes Outlook/OneDrive/Calendar/Teams via Graph.
|
||||
|
||||
- id: atlassian
|
||||
label: Atlassian (Jira + Confluence)
|
||||
label: Atlassian (Jira + Confluence) — upstream mirror source
|
||||
category: saas
|
||||
kind: oauth
|
||||
auth: oauth2
|
||||
kind: api-key
|
||||
auth: api_key
|
||||
endpoint: https://api.atlassian.com
|
||||
surfaces: [mcp, theia]
|
||||
status: declared
|
||||
scopes: [read:jira-work, write:jira-work, read:confluence-content.all, offline_access]
|
||||
scopes: [read:jira-work, read:confluence-content.all, offline_access]
|
||||
host: null
|
||||
notes: Atlassian OAuth 2.0 (3LO) app; Jira REST v3 + Confluence REST v2.
|
||||
notes: |
|
||||
Uppstream source for the mirror pipeline (Jira → Plane, Confluence → Outline).
|
||||
Uses API-token auth (email + token), not OAuth, for server-side sync.
|
||||
Configured in .aurelio/mirrors/sync-mirrors.yaml under targets.plane/outline.
|
||||
Fetch tokens from Vaultwarden; env vars JIRA_API_TOKEN/JIRA_USER_EMAIL and
|
||||
CONFLUENCE_API_TOKEN/CONFLUENCE_USER_EMAIL.
|
||||
|
||||
# ── Self-hosted mirrors (receive upstream via sync-mirrors.py) ────────────
|
||||
- id: plane
|
||||
label: Plane (issue tracker — Jira mirror)
|
||||
category: devops
|
||||
kind: api-key
|
||||
auth: api_key
|
||||
endpoint: https://plane.portugalfuturista.org
|
||||
surfaces: [mcp, theia, portal]
|
||||
status: live
|
||||
scopes: []
|
||||
host: null
|
||||
notes: |
|
||||
Self-hosted Plane. Mirror target for Jira issues via scripts/sync-mirrors.py --sync plane.
|
||||
Workspace: savearth. Token env: PLANE_API_TOKEN. REST API under /api/v1/.
|
||||
|
||||
- id: outline
|
||||
label: Outline (docs — Confluence mirror)
|
||||
category: devops
|
||||
kind: api-key
|
||||
auth: api_key
|
||||
endpoint: https://docs.portugalfuturista.org
|
||||
surfaces: [mcp, theia, portal]
|
||||
status: live
|
||||
scopes: []
|
||||
host: null
|
||||
notes: |
|
||||
Self-hosted Outline. Mirror target for Confluence pages via scripts/sync-mirrors.py --sync outline.
|
||||
Collection: "Savearth Mirror". Token env: OUTLINE_API_TOKEN. REST API under /api/.
|
||||
|
||||
# ── Creative / CAD hosts (sidecar bridges, NOT VSIXs) ─────────────────────
|
||||
- id: solidworks
|
||||
|
|
|
|||
185
.aurelio/garden/agent-garden/google-cloud.yaml
Normal file
185
.aurelio/garden/agent-garden/google-cloud.yaml
Normal file
|
|
@ -0,0 +1,185 @@
|
|||
# PF Agent Garden — Google Cloud
|
||||
#
|
||||
# Agents and agentic services that run on GCP or call Google Cloud / Workspace APIs.
|
||||
# Distinct from the generic `gemini` chat provider.
|
||||
|
||||
platform: google-cloud
|
||||
realm_default: europe-west1
|
||||
workload_identity: required
|
||||
|
||||
agents:
|
||||
- id: gcp-deployer
|
||||
name: GCP Deployer
|
||||
description: Provisions and mutates GCP resources (GKE, Cloud Run, Cloud Functions, Cloud Storage, Pub/Sub, BigQuery, Cloud SQL, Secret Manager) from declarative specs.
|
||||
kind: infrastructure
|
||||
surfaces:
|
||||
- dirac
|
||||
- portal
|
||||
triggers:
|
||||
- user-command
|
||||
- git-push
|
||||
inputs:
|
||||
- terraform
|
||||
- gcloud-yaml
|
||||
- openapi
|
||||
outputs:
|
||||
- deployed-url
|
||||
- resource-manifest
|
||||
required_apis:
|
||||
- deploymentmanager.googleapis.com
|
||||
- cloudbuild.googleapis.com
|
||||
- run.googleapis.com
|
||||
- container.googleapis.com
|
||||
|
||||
- id: vertex-coder
|
||||
name: Vertex Coder
|
||||
description: Coding agent backed by Vertex AI Gemini. Reads/writes code in a GKE/Cloud Run sandbox and streams results through Dirac / MCP.
|
||||
kind: coding
|
||||
surfaces:
|
||||
- dirac
|
||||
- mcp
|
||||
model: gemini-2.5-pro-preview-06-05
|
||||
runtime: cloud-run
|
||||
triggers:
|
||||
- dirac-task
|
||||
- mcp-tool-call
|
||||
inputs:
|
||||
- repository
|
||||
- prompt
|
||||
outputs:
|
||||
- diff
|
||||
- test-results
|
||||
|
||||
- id: gcp-cost-guardian
|
||||
name: GCP Cost Guardian
|
||||
description: Monitors billing, budgets, and resource labels; alerts on anomalies and suggests rightsizing.
|
||||
kind: observability
|
||||
surfaces:
|
||||
- portal
|
||||
- mcp
|
||||
triggers:
|
||||
- schedule
|
||||
- billing-alert
|
||||
inputs:
|
||||
- billing-export
|
||||
- resource-labels
|
||||
outputs:
|
||||
- alert
|
||||
- recommendation
|
||||
required_roles:
|
||||
- roles/billing.viewer
|
||||
- roles/monitoring.viewer
|
||||
|
||||
- id: workspace-delegator
|
||||
name: Workspace Delegator
|
||||
description: Calls Google Workspace APIs (Directory, Gmail, Calendar, Drive) via domain-wide delegation. Useful for tenant automation and mail/calendar agents.
|
||||
kind: integration
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
triggers:
|
||||
- user-command
|
||||
- workflow
|
||||
inputs:
|
||||
- workspace-scope
|
||||
- service-account
|
||||
outputs:
|
||||
- api-result
|
||||
- event
|
||||
requires_domain_wide_delegation: true
|
||||
scopes:
|
||||
- https://www.googleapis.com/auth/admin.directory.user.readonly
|
||||
- https://www.googleapis.com/auth/calendar
|
||||
- https://www.googleapis.com/auth/gmail.modify
|
||||
- https://www.googleapis.com/auth/drive
|
||||
|
||||
- id: pubsub-agent-router
|
||||
name: Pub/Sub Agent Router
|
||||
description: Receives events from Pub/Sub and dispatches them to other PF agents running on Cloud Run, GKE, or via MCP.
|
||||
kind: orchestration
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
triggers:
|
||||
- pubsub-message
|
||||
inputs:
|
||||
- pubsub-payload
|
||||
- routing-key
|
||||
outputs:
|
||||
- agent-invocation
|
||||
required_apis:
|
||||
- pubsub.googleapis.com
|
||||
- run.googleapis.com
|
||||
|
||||
- id: bq-memory-analyst
|
||||
name: BigQuery Memory Analyst
|
||||
description: Queries structured agent memory and telemetry stored in BigQuery; returns synthesized answers with citations.
|
||||
kind: memory
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
model: gemini-2.5-flash-preview-06-05
|
||||
triggers:
|
||||
- user-query
|
||||
- schedule
|
||||
inputs:
|
||||
- bigquery-dataset
|
||||
- sql-question
|
||||
outputs:
|
||||
- sql-query
|
||||
- result-summary
|
||||
required_roles:
|
||||
- roles/bigquery.dataViewer
|
||||
- roles/bigquery.jobUser
|
||||
|
||||
- id: gcs-artifact-librarian
|
||||
name: GCS Artifact Librarian
|
||||
description: Indexes and retrieves artifacts from Cloud Storage buckets; supports code artifacts, documents, and media.
|
||||
kind: memory
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
triggers:
|
||||
- user-query
|
||||
- event
|
||||
inputs:
|
||||
- bucket-prefix
|
||||
- metadata
|
||||
outputs:
|
||||
- signed-url
|
||||
- object-list
|
||||
required_roles:
|
||||
- roles/storage.objectViewer
|
||||
|
||||
- id: secret-bootstrapper
|
||||
name: Secret Bootstrapper
|
||||
description: Loads runtime secrets from Secret Manager into agent workloads; never logs values and rotates on schedule.
|
||||
kind: security
|
||||
surfaces:
|
||||
- mcp
|
||||
- dirac
|
||||
triggers:
|
||||
- deployment
|
||||
- rotation-schedule
|
||||
inputs:
|
||||
- secret-name
|
||||
- workload-identity
|
||||
outputs:
|
||||
- mounted-secret
|
||||
required_apis:
|
||||
- secretmanager.googleapis.com
|
||||
|
||||
runtimes:
|
||||
- name: cloud-run
|
||||
use_for: stateless-http
|
||||
scaling: request-driven
|
||||
- name: gke-autopilot
|
||||
use_for: long-running-workloads
|
||||
scaling: workload-driven
|
||||
- name: cloud-functions
|
||||
use_for: event-driven-functions
|
||||
|
||||
notes:
|
||||
- All agents default to `europe-west1` unless EU residency or latency requires otherwise.
|
||||
- Production agents must use Workload Identity or dedicated service accounts; no downloaded keys.
|
||||
- Agents calling Workspace APIs require domain-wide delegation registered in the Google Admin console.
|
||||
122
.aurelio/garden/model-garden/google-cloud.yaml
Normal file
122
.aurelio/garden/model-garden/google-cloud.yaml
Normal file
|
|
@ -0,0 +1,122 @@
|
|||
# PF Model Garden — Google Cloud (Vertex AI)
|
||||
#
|
||||
# Enterprise Google Cloud models exposed through the Aurelio ecosystem.
|
||||
# These are distinct from the consumer Gemini (AI Studio) provider.
|
||||
|
||||
provider: google-cloud
|
||||
label: Google Cloud Vertex AI
|
||||
protocol: gemini
|
||||
auth:
|
||||
kind: service_account
|
||||
env: GOOGLE_APPLICATION_CREDENTIALS
|
||||
fallback_env: GOOGLE_CLOUD_API_KEY
|
||||
region_default: europe-west1
|
||||
|
||||
models:
|
||||
- id: gemini-2.5-pro-preview-06-05
|
||||
name: Gemini 2.5 Pro
|
||||
kind: chat
|
||||
publisher: google
|
||||
context_window: 1000000
|
||||
surfaces:
|
||||
- dirac
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
input_per_1m_tokens: 1.25
|
||||
output_per_1m_tokens: 10.00
|
||||
currency: USD
|
||||
notes: Approximate; region and batch mode affect rate. Check cloud.google.com/vertex-ai/generative-ai/pricing.
|
||||
|
||||
- id: gemini-2.5-flash-preview-06-05
|
||||
name: Gemini 2.5 Flash
|
||||
kind: chat
|
||||
publisher: google
|
||||
context_window: 1000000
|
||||
surfaces:
|
||||
- dirac
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
input_per_1m_tokens: 0.15
|
||||
output_per_1m_tokens: 0.60
|
||||
currency: USD
|
||||
notes: Fast, low-cost; good for agentic loops and classification.
|
||||
|
||||
- id: gemini-2.0-flash-001
|
||||
name: Gemini 2.0 Flash
|
||||
kind: chat
|
||||
publisher: google
|
||||
context_window: 1000000
|
||||
surfaces:
|
||||
- dirac
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
input_per_1m_tokens: 0.10
|
||||
output_per_1m_tokens: 0.40
|
||||
currency: USD
|
||||
notes: Stable GA model.
|
||||
|
||||
- id: imagen-3-0-generate-001
|
||||
name: Imagen 3
|
||||
kind: image
|
||||
publisher: google
|
||||
context_window: null
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
per_image: 0.04
|
||||
currency: USD
|
||||
notes: Per image at 1024x1024; higher resolutions cost more.
|
||||
|
||||
- id: veo-2-0-generate-001
|
||||
name: Veo 2
|
||||
kind: video
|
||||
publisher: google
|
||||
context_window: null
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
per_second: 0.05
|
||||
currency: USD
|
||||
notes: Per generated video second; minimum length applies.
|
||||
|
||||
- id: text-embedding-004
|
||||
name: Text Embedding 004
|
||||
kind: embedding
|
||||
publisher: google
|
||||
context_window: null
|
||||
dimensions: 2048
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
per_1m_tokens: 0.10
|
||||
currency: USD
|
||||
notes: Input-only embedding model.
|
||||
|
||||
- id: multimodalembedding@001
|
||||
name: Multimodal Embedding
|
||||
kind: embedding
|
||||
publisher: google
|
||||
context_window: null
|
||||
dimensions: [128, 256, 512, 1408]
|
||||
surfaces:
|
||||
- mcp
|
||||
- portal
|
||||
pricing_proxy:
|
||||
per_1m_tokens: 0.20
|
||||
currency: USD
|
||||
notes: Mixed image+text embeddings; dimensions selectable.
|
||||
|
||||
endpoints:
|
||||
chat_template: "https://{region}-aiplatform.googleapis.com/v1/projects/{project}/locations/{region}/publishers/google/models/{model_id}:generateContent"
|
||||
stream_template: "https://{region}-aiplatform.googleapis.com/v1/projects/{project}/locations/{region}/publishers/google/models/{model_id}:streamGenerateContent"
|
||||
|
||||
notes:
|
||||
- Model availability and pricing vary by region; verify before production use.
|
||||
- Use service-account auth in production; API-key fallback is for quick tests only.
|
||||
- The separate `gemini` provider in .aurelio/providers/registry.yaml targets AI Studio (consumer), not Vertex AI.
|
||||
|
|
@ -148,10 +148,32 @@
|
|||
"mcp-remote",
|
||||
"https://stitch.googleapis.com/mcp",
|
||||
"--header",
|
||||
"X-Goog-Api-Key: AQ.Ab8RN6KphA2R_SVCtr5-Sy1PAjPzV6lZnLecLdHjQbpdo7nGVA"
|
||||
"X-Goog-Api-Key: AQ.Ab8...GVA"
|
||||
],
|
||||
"env": {},
|
||||
"_disabled": false
|
||||
},
|
||||
"muscriptor-mcp": {
|
||||
"command": "npx",
|
||||
"args": [
|
||||
"-y",
|
||||
"mcp-remote",
|
||||
"http://192.168.0.104:8088/sse"
|
||||
],
|
||||
"_disabled": false,
|
||||
"_note": "Music transcription MCP — audio → MIDI via MuScriptor. Runs on Gigabyte (GPU) port 8088. [LAN URL]",
|
||||
"disabledTools": []
|
||||
},
|
||||
"music-mcp": {
|
||||
"command": "npx",
|
||||
"args": [
|
||||
"-y",
|
||||
"mcp-remote",
|
||||
"http://192.168.0.104:8089/sse"
|
||||
],
|
||||
"_disabled": false,
|
||||
"_note": "Music stack MCP — Navidrome + Lidarr + Last.fm + MusicBrainz + MuScriptor. Debug mode, daily brief, MIDI conversion. Runs on Gigabyte port 8089. [LAN URL]",
|
||||
"disabledTools": []
|
||||
}
|
||||
}
|
||||
}
|
||||
156
.aurelio/mirrors/README.md
Normal file
156
.aurelio/mirrors/README.md
Normal file
|
|
@ -0,0 +1,156 @@
|
|||
# Mirror Pipeline — Upstream Replication into Self-Hosted Mirrors
|
||||
|
||||
Constant, one-way replication of Savearth's upstream sources into Portugal
|
||||
Futurista's self-hosted mirrors. The agent fleet operates on the mirrors; the
|
||||
originals stay authoritative.
|
||||
|
||||
```
|
||||
UPSTREAM (source of truth) MIRROR (agents work here)
|
||||
────────────────────────── ─────────────────────────
|
||||
GitHub (SavearthTech/*) ──────► Forgejo code.portugalfuturista.org
|
||||
Jira Cloud ──────► Plane plane.portugalfuturista.org
|
||||
Confluence Cloud ──────► Outline docs.portugalfuturista.org
|
||||
│
|
||||
sync outcomes
|
||||
▼
|
||||
trajectory-rewards/mirror-sync.jsonl
|
||||
│
|
||||
▼
|
||||
GRPO/DPO RL pipeline (rl-agent-trajectory-training)
|
||||
```
|
||||
|
||||
## Architecture
|
||||
|
||||
Single source of truth, matching the repo's established provider/connector
|
||||
pattern:
|
||||
|
||||
| File | Role |
|
||||
|------|------|
|
||||
| `.aurelio/mirrors/sync-mirrors.yaml` | **Edit this** — declares sources, targets, auth refs |
|
||||
| `.aurelio/mirrors/dist/mirrors.catalog.json` | **Generated** — catalog for downstream surfaces |
|
||||
| `.aurelio/mirrors/state/last-sync.json` | **Generated** — last sync summary (for drift detection) |
|
||||
| `scripts/sync-mirrors.py` | The engine: three sync modes |
|
||||
| `.forgejo/workflows/mirror-sync.yml` | CI: runs every 15 min |
|
||||
| `.forgejo/workflows/ci.yml` | CI: `--check` guard on every push |
|
||||
|
||||
## Usage
|
||||
|
||||
### 1. Provision tokens
|
||||
|
||||
All tokens are API tokens/PATs. Fetch from Vaultwarden, export as env vars:
|
||||
|
||||
```bash
|
||||
export FORGEJO_MIRROR_TOKEN=... # Forgejo admin (manage_repo, write:org)
|
||||
export GITHUB_MIRROR_TOKEN=... # GitHub PAT classic (repo, read:org)
|
||||
export PLANE_API_TOKEN=... # Plane personal API token
|
||||
export OUTLINE_API_TOKEN=... # Outline API token (read+write)
|
||||
export JIRA_API_TOKEN=... # Atlassian API token
|
||||
export JIRA_USER_EMAIL=... # Atlassian account email
|
||||
export CONFLUENCE_API_TOKEN=... # Same Atlassian account
|
||||
export CONFLUENCE_USER_EMAIL=...
|
||||
```
|
||||
|
||||
See `.env.example` for generation instructions per service.
|
||||
|
||||
### 2. Configure upstream endpoints
|
||||
|
||||
Edit `.aurelio/mirrors/sync-mirrors.yaml`:
|
||||
|
||||
```yaml
|
||||
targets:
|
||||
plane:
|
||||
sources:
|
||||
- id: savearth-jira
|
||||
endpoint: https://savearth.atlassian.net # ← set this
|
||||
projects: [SAVE, IOT, HW] # ← or null for all
|
||||
```
|
||||
|
||||
The Jira and Confluence `endpoint` fields are `null` by default — the pipeline
|
||||
skips them until you set the real URL.
|
||||
|
||||
### 3. Run the sync
|
||||
|
||||
```bash
|
||||
# One-time / manual:
|
||||
python3 scripts/sync-mirrors.py --sync # all targets
|
||||
python3 scripts/sync-mirrors.py --sync forgejo # one target
|
||||
python3 scripts/sync-mirrors.py --sync --dry-run # preview
|
||||
|
||||
# Regenerate catalog after editing the YAML:
|
||||
python3 scripts/sync-mirrors.py --write-in-place
|
||||
|
||||
# CI guard:
|
||||
python3 scripts/sync-mirrors.py --check # exit 1 if dist stale
|
||||
```
|
||||
|
||||
### 4. Constant sync (CI)
|
||||
|
||||
`.forgejo/workflows/mirror-sync.yml` runs every 15 minutes. Tokens must be set
|
||||
as Forgejo repository secrets:
|
||||
|
||||
```
|
||||
Settings → Secrets → Actions → New Secret:
|
||||
FORGEJO_MIRROR_TOKEN, GITHUB_MIRROR_TOKEN, PLANE_API_TOKEN,
|
||||
OUTLINE_API_TOKEN, JIRA_API_TOKEN, JIRA_USER_EMAIL,
|
||||
CONFLUENCE_API_TOKEN, CONFLUENCE_USER_EMAIL
|
||||
```
|
||||
|
||||
The workflow commits `last-sync.json` + reward signals back to the repo.
|
||||
|
||||
## What each mirror does
|
||||
|
||||
### GitHub → Forgejo (code)
|
||||
|
||||
- Lists all repos in the `SavearthTech` org (+ personal matching repos).
|
||||
- Creates a Forgejo migration with `mirror: true` — Forgejo polls upstream
|
||||
on its own schedule (`native_interval: 8h`) after the initial clone.
|
||||
- Skips forks and archived repos by default.
|
||||
- Preserves private status, description, topics→labels.
|
||||
|
||||
### Jira → Plane (issues)
|
||||
|
||||
- Lists Jira projects, creates matching Plane projects (by key/name).
|
||||
- Imports issues with field mapping (summary→name, description→description_html,
|
||||
priority, status, assignee, labels).
|
||||
- Converts Atlassian Document Format (ADF) to HTML for Plane.
|
||||
- Idempotent: skips issues that already exist by name match.
|
||||
|
||||
### Confluence → Outline (docs)
|
||||
|
||||
- Lists Confluence spaces, creates an Outline collection ("Savearth Mirror").
|
||||
- Imports pages with `[SPACE_KEY] title` naming.
|
||||
- Converts Confluence storage format (XHTML) to Markdown.
|
||||
- Handles headings, code blocks, lists, links, bold/italic.
|
||||
- Deduplicates by title within the collection.
|
||||
|
||||
## RL feedback loop
|
||||
|
||||
Each sync run emits reward signals to
|
||||
`.aurelio/brain/trajectory-rewards/mirror-sync.jsonl`:
|
||||
|
||||
| Signal | Reward | When |
|
||||
|--------|--------|------|
|
||||
| `sync_hit` | +0.1 | Target synced cleanly (no errors) |
|
||||
| `sync_miss` | -0.2 | Target had sync errors (drift/staleness) |
|
||||
| `upstream_overwrite` | -0.5 | Agent's mirror change was overwritten by upstream sync |
|
||||
|
||||
These feed into the GRPO/DPO pipeline (see `rl-agent-trajectory-training`
|
||||
skill). The reward shaping is configured in `sync-mirrors.yaml` under
|
||||
`rl_feedback.signals`.
|
||||
|
||||
## Direction: read-only mirror
|
||||
|
||||
Upstream is the source of truth. Agents work on the Forgejo/Plane/Outline
|
||||
copies. Changes do NOT propagate back to GitHub/Jira/Confluence. This is by
|
||||
design — the pipeline is structured so bidirectional write-back can be added
|
||||
per-target later without rewriting the engine.
|
||||
|
||||
## Troubleshooting
|
||||
|
||||
| Symptom | Fix |
|
||||
|---------|-----|
|
||||
| `missing env FORGEJO_MIRROR_TOKEN` | Export the token (see step 1) |
|
||||
| `Jira endpoint is null` | Set `endpoint:` in `sync-mirrors.yaml` |
|
||||
| `0 repos visible` from GitHub | Token lacks `read:org` or repos are private — check PAT scopes |
|
||||
| Plane 401/403 | Regenerate API token in Plane settings |
|
||||
| Outline OAuth error | Outline needs `ENABLE_LOCAL_AUTH=true` or Authentik OIDC (see fleet/identity.md) |
|
||||
4
.aurelio/mirrors/state/last-sync.json
Normal file
4
.aurelio/mirrors/state/last-sync.json
Normal file
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"timestamp": "2026-07-30T22:58:11.084761+00:00",
|
||||
"summaries": []
|
||||
}
|
||||
159
.aurelio/mirrors/sync-mirrors.yaml
Normal file
159
.aurelio/mirrors/sync-mirrors.yaml
Normal file
|
|
@ -0,0 +1,159 @@
|
|||
# Mirror Pipeline — single source of truth
|
||||
#
|
||||
# Declares how upstream sources (GitHub, Jira, Confluence) replicate into the
|
||||
# self-hosted mirrors (Forgejo, Plane, Outline). All downstream surfaces derive
|
||||
# from this file via scripts/sync-mirrors.py.
|
||||
#
|
||||
# ONE place to edit. mirrors/dist/*.json are GENERATED (do not edit).
|
||||
#
|
||||
# Direction: upstream → mirror (read-only by default; write-back is a future
|
||||
# toggle per-target). The agent fleet operates on the mirrors; the originals
|
||||
# stay authoritative.
|
||||
#
|
||||
# Regenerate after editing:
|
||||
# python3 scripts/sync-mirrors.py --write-in-place # regenerate dist/
|
||||
# python3 scripts/sync-mirrors.py --check # CI guard: exit 1 if stale
|
||||
# python3 scripts/sync-mirrors.py --sync # run all mirror syncs
|
||||
# python3 scripts/sync-mirrors.py --sync forgejo # one target
|
||||
#
|
||||
# Auth: tokens are NEVER in this file. They live in the environment (.env or
|
||||
# shell env), referenced by the `token_env` key. Fetch from Vaultwarden.
|
||||
|
||||
version: 1
|
||||
|
||||
# ── Global defaults ────────────────────────────────────────────────────────
|
||||
defaults:
|
||||
# Sync cadence (minutes). Overridden by cron job schedule.
|
||||
interval_minutes: 15
|
||||
# Fail the whole run if any single source errors? false = log + continue.
|
||||
fail_fast: false
|
||||
# User-agent for upstream API calls.
|
||||
user_agent: "replica-omnisciente-mirror/1.0"
|
||||
|
||||
# ── Mirror targets ─────────────────────────────────────────────────────────
|
||||
# Each target is a self-hosted service that RECEIVES data from one or more
|
||||
# upstream sources. A target may have multiple sources (e.g. Forgejo mirrors
|
||||
# both GitHub orgs and individual repos).
|
||||
|
||||
targets:
|
||||
|
||||
# ── Forgejo: code mirror (GitHub → Forgejo) ────────────────────────────
|
||||
forgejo:
|
||||
kind: git-mirror
|
||||
label: "Forgejo (code mirror)"
|
||||
endpoint: https://code.portugalfuturista.org
|
||||
# Forgejo API token (admin or org-owner scope). ENV var name only.
|
||||
token_env: FORGEJO_MIRROR_TOKEN
|
||||
# Default owner/org for mirrored repos if the upstream org doesn't exist.
|
||||
default_owner: SavearthTech
|
||||
# Mirror interval in Forgejo's own scheduler (for native push mirrors).
|
||||
native_interval: 8h
|
||||
sources:
|
||||
- id: savearthtech-github
|
||||
label: "SavearthTech GitHub org"
|
||||
type: github-org
|
||||
# GitHub org name to mirror ALL public+private repos from.
|
||||
org: SavearthTech
|
||||
# GitHub PAT (classic, read scope on repo + org). ENV var name only.
|
||||
token_env: GITHUB_MIRROR_TOKEN
|
||||
# Clone over SSH for private repos (key must be in agent).
|
||||
clone_transport: https
|
||||
# Include forked repos? false = skip forks.
|
||||
include_forks: false
|
||||
# Map GitHub topics → Forgejo labels (optional).
|
||||
topic_labels: true
|
||||
status: declared
|
||||
|
||||
- id: savearth-personal-github
|
||||
label: "Personal Savearth repos (fabiorafaelcoutada)"
|
||||
type: github-user
|
||||
user: fabiorafaelcoutada
|
||||
token_env: GITHUB_MIRROR_TOKEN
|
||||
# Only mirror repos matching these patterns (glob).
|
||||
include_patterns:
|
||||
- "savearth*"
|
||||
- "aws-iot-core*"
|
||||
- "hardware-devices*"
|
||||
- "agent-firmware*"
|
||||
exclude_patterns:
|
||||
- "*.github.io"
|
||||
clone_transport: https
|
||||
status: declared
|
||||
|
||||
# ── Plane: issue tracker mirror (Jira → Plane) ─────────────────────────
|
||||
plane:
|
||||
kind: issue-mirror
|
||||
label: "Plane (issue tracker mirror)"
|
||||
endpoint: https://plane.portugalfuturista.org
|
||||
token_env: PLANE_API_TOKEN
|
||||
# Plane workspace slug (in the URL). Create if it doesn't exist.
|
||||
workspace_slug: savearth
|
||||
sources:
|
||||
- id: savearth-jira
|
||||
label: "Savearth Jira Cloud"
|
||||
type: jira-cloud
|
||||
# Jira Cloud base URL (e.g. https://savearth.atlassian.net).
|
||||
# Leave null if Savearth doesn't use Jira Cloud — the pipeline will
|
||||
# skip this source and report it.
|
||||
endpoint: null
|
||||
token_env: JIRA_API_TOKEN
|
||||
# Jira user email (for API token auth). ENV var name only.
|
||||
email_env: JIRA_USER_EMAIL
|
||||
# Jira projects to mirror (keys, e.g. SAVE, IOT, HW). null = all.
|
||||
projects: null
|
||||
# Map Jira → Plane fields. Plane uses: name, description_html,
|
||||
# state (group), priority (urgency), assignee, labels, cycle.
|
||||
field_map:
|
||||
summary: name
|
||||
description: description_html
|
||||
status: state
|
||||
priority: priority
|
||||
assignee: assignee
|
||||
labels: labels
|
||||
issuetype: type
|
||||
status: declared
|
||||
|
||||
# ── Outline: documentation mirror (Confluence → Outline) ───────────────
|
||||
outline:
|
||||
kind: doc-mirror
|
||||
label: "Outline (documentation mirror)"
|
||||
endpoint: https://docs.portugalfuturista.org
|
||||
token_env: OUTLINE_API_TOKEN
|
||||
# Outline collection to mirror into (created if absent).
|
||||
collection_name: Savearth Mirror
|
||||
sources:
|
||||
- id: savearth-confluence
|
||||
label: "Savearth Confluence Cloud"
|
||||
type: confluence-cloud
|
||||
# Confluence Cloud base URL (e.g. https://savearth.atlassian.net/wiki).
|
||||
endpoint: null
|
||||
token_env: CONFLUENCE_API_TOKEN
|
||||
email_env: CONFLUENCE_USER_EMAIL
|
||||
# Space keys to mirror (e.g. SAVE, ENG, HW). null = all.
|
||||
spaces: null
|
||||
# Convert Confluence storage format → Markdown for Outline.
|
||||
format: markdown
|
||||
# Include attachments? (binary download + re-upload to Outline).
|
||||
include_attachments: true
|
||||
status: declared
|
||||
|
||||
# ── RL feedback loop ──────────────────────────────────────────────────────
|
||||
# Mirror sync outcomes feed into the RL trajectory pipeline as reward signals.
|
||||
# When an agent operates on a mirror and the sync detects drift (upstream
|
||||
# changed but mirror lagged, or an agent's change diverged from upstream),
|
||||
# that becomes a negative reward signal for the agent's trajectory.
|
||||
#
|
||||
# See scripts/sync-mirrors.py --emit-trajectory-rewards and the
|
||||
# rl-agent-trajectory-training skill for the consumption side.
|
||||
rl_feedback:
|
||||
enabled: true
|
||||
# Where to write reward signals (trajectory.jsonl format, appended).
|
||||
output: .aurelio/brain/trajectory-rewards/mirror-sync.jsonl
|
||||
# Reward shaping: how much each signal type contributes.
|
||||
signals:
|
||||
# Agent worked on a mirror that was in sync with upstream (positive).
|
||||
sync_hit: 0.1
|
||||
# Agent worked on a mirror that had drifted from upstream (negative).
|
||||
sync_miss: -0.2
|
||||
# Agent's mirror change was overwritten by an upstream sync (negative).
|
||||
upstream_overwrite: -0.5
|
||||
|
|
@ -1,576 +1,86 @@
|
|||
# Aurelio Omnibrain — Execution Tracker
|
||||
# Aurelio OmniBrain — Execution Tracker
|
||||
|
||||
> **Generated:** 2026-07-13 · **Source:** `~/.hermes/plans/2026-07-12_aurelio-omnibrain-master-plan.md`
|
||||
> **Status:** 🔴 NOT STARTED — zero execution as of today.
|
||||
**Date:** 2026-07-13
|
||||
**Status:** Draft (scaffold — confirm objectives & priorities)
|
||||
**Owner:** Aurélio / Portugal Futurista
|
||||
**Scope:** `replica-omnisciente` orchestration brain across realms, wired to
|
||||
`mycelium` (XSD→codegen), `aurelio-theia` (IDE), `universalisos`, and `gbrain`.
|
||||
|
||||
> Derived from `mycelium/.aurelio/plans/2026-07-12_portugal-futurista-12month-master-plan.md`
|
||||
> and `2026-07-12_mycelium-agent-blocks-web-presence.md`. This file is a
|
||||
> **starting scaffold** — replace the task list with the real program once
|
||||
> confirmed.
|
||||
|
||||
---
|
||||
|
||||
## 1. Vision
|
||||
|
||||
Import session history, plans, memory, and artifacts from **every coding-agent CLI/TUI/IDE** on the market into a unified **Aurelio GBrain** engine, and expose that brain as the **single knowledge backend** for the entire Aurélio fleet:
|
||||
|
||||
| Surface | How it talks to GBrain |
|
||||
|---------|----------------------|
|
||||
| **Theia IDE** | HTTP via `GBrainClient` (Node.js backend) |
|
||||
| **VS Code extension** | HTTP via `GBrainClient` (TypeScript) |
|
||||
| **TUI / CLI** | HTTP via `GBrainClient` (Node.js) |
|
||||
| **Hermes Agent** | MCP server (stdio) or HTTP |
|
||||
|
||||
**North star:** Every coding session you've ever had — Kimi, Claude, Hermes, OpenCode, Copilot, Gemini, ZAI, ZCode, MiMo, Qwen, Antigravity — searchable from one place, available in every IDE.
|
||||
|
||||
---
|
||||
|
||||
## 2. Current State (Verified 2026-07-13)
|
||||
|
||||
### Brain Store
|
||||
```
|
||||
~/workspace/replica-omnisciente/.aurelio/brain/
|
||||
```
|
||||
| Source | Sessions | Status |
|
||||
|--------|----------|--------|
|
||||
| **kimi** | 539 | ✅ Imported |
|
||||
| **claude** | 127 | ✅ Imported |
|
||||
| **antigravity** | 3 | ✅ Imported |
|
||||
| **mimocode** | 2 | ✅ Imported |
|
||||
| **hermes** | 1 | ✅ Imported |
|
||||
| **legacy UUID** | ~127 | ⚠️ Pre-dating the importer (no `session-<source>-` prefix) |
|
||||
| **kimi-code** | 0 | ❌ No adapter |
|
||||
| **antigravity-ide** | 0 | ❌ No adapter |
|
||||
| **gemini** | 0 | ❌ No adapter |
|
||||
| **copilot** | 0 | ❌ No adapter |
|
||||
| **opencode** | 0 | ❌ No adapter |
|
||||
| **zai** | 0 | ❌ No adapter |
|
||||
| **zcode** | 0 | ❌ No adapter |
|
||||
| **qwen-code** | 0 | ❌ Not imported yet (adapter exists) |
|
||||
| **Total** | **~800** | |
|
||||
|
||||
### Code Components
|
||||
|
||||
| Component | Path | Reality |
|
||||
|-----------|------|---------|
|
||||
| **Agent importer** | `scripts/sync-agents-to-brain.py` | ✅ Exists. Monolithic `adapters.py` (not yet split into package). 6 adapters. Idempotent. |
|
||||
| **Adapters** | `scripts/agent_importers/adapters.py` | 🟡 Single file, not a package. Needs refactor before adding 5+ new adapters. |
|
||||
| **Normalized contract** | `scripts/agent_importers/engine.py` | ✅ `NormalizedSession` / `NormalizedMessage` dataclasses. |
|
||||
| **GBrain engine** | `aurelio-gbrain/` | 🔴 **Empty directory.** No code. |
|
||||
| **Theia extension** | `aurelio-theia/AurelioTheia/` | 🔴 **Thin scaffold.** `aureilio-backend-module.ts` (hardcoded endpoints), `aurelio-frontend-module.ts` (50+ `console.log` stubs). No widgets. |
|
||||
| **Gabinete Hub** | CT 208 (192.168.0.38) | 🟡 Exists as deployment target. No GBrain integration. |
|
||||
| **Provider registry** | `.aurelio/providers/registry.yaml` | ✅ Covers 17+ providers. |
|
||||
|
||||
### Source Directory Map
|
||||
|
||||
| # | Agent / CLI / TUI | Source Directory | Adapter? | Sessions |
|
||||
|---|-------------------|------------------|----------|----------|
|
||||
| 1 | Kimi | `~/.kimi/sessions/` | ✅ | 539 |
|
||||
| 2 | Kimi Code | `~/.kimi-code/` | ❌ NEW | ? |
|
||||
| 3 | Hermes | `~/.hermes/sessions/` | ✅ | 1 |
|
||||
| 4 | Claude Code | `~/.claude/projects/` | ✅ | 127 |
|
||||
| 5 | Antigravity IDE | `~/.antigravity-ide/` | ❌ NEW | ? |
|
||||
| 6 | Antigravity CLI | `~/.gemini/antigravity-cli/` | ✅ | 3 |
|
||||
| 7 | Gemini | `~/.gemini/` | ❌ NEW | ? |
|
||||
| 8 | Copilot | `~/.copilot/` | ❌ NEW | ? |
|
||||
| 9 | OpenCode | `~/.opencode/` | ❌ NEW | ? |
|
||||
| 10 | ZAI | `~/.zai/` | ❌ NEW | ? |
|
||||
| 11 | ZCode | `~/.zcode/` | ❌ NEW | ? |
|
||||
| 12 | MiMo Code | `~/.mimocode/` | ✅ | 2 |
|
||||
| 13 | Qwen Code | `~/.qwen-code/` | ✅ | 0 |
|
||||
|
||||
---
|
||||
|
||||
## 3. Architecture
|
||||
|
||||
```
|
||||
┌─────────────────────────────────────────────┐
|
||||
│ AURELIO FLEET │
|
||||
│ │
|
||||
│ ┌─────────┐ ┌─────────┐ ┌─────────┐ │
|
||||
│ │ Theia │ │ VS Code │ │ TUI / │ │
|
||||
│ │ IDE │ │ ext. │ │ CLI │ │
|
||||
│ └────┬────┘ └────┬────┘ └────┬────┘ │
|
||||
│ │ │ │ │
|
||||
│ └───────────┼────────────┘ │
|
||||
│ │ HTTP │
|
||||
│ ▼ │
|
||||
│ ┌────────────────────────────────────┐ │
|
||||
│ │ AURELIO GBRAIN ENGINE │ │
|
||||
│ │ ┌──────────┐ ┌────────────────┐ │ │
|
||||
│ │ │ SQLite │ │ HTTP API │ │ │
|
||||
│ │ │ FTS5 │ │ /api/brain/* │ │ │
|
||||
│ │ │ Index │ │ (Express) │ │ │
|
||||
│ │ └──────────┘ └────────────────┘ │ │
|
||||
│ │ ┌──────────┐ ┌────────────────┐ │ │
|
||||
│ │ │ Dream │ │ MCP Server │ │ │
|
||||
│ │ │ Cycle │ │ (stdio) │ │ │
|
||||
│ │ └──────────┘ └────────────────┘ │ │
|
||||
│ └────────────────────────────────────┘ │
|
||||
│ ▲ │
|
||||
│ │ │
|
||||
│ ┌────────────────────────────────────┐ │
|
||||
│ │ AGENT IMPORTERS │ │
|
||||
│ │ sync-agents-to-brain.py │ │
|
||||
│ │ ┌─────┐ ┌──────┐ ┌───────┐ │ │
|
||||
│ │ │ kimi│ │claude│ │opencode│ ... │ │
|
||||
│ │ │code │ │ code │ │ │ │ │
|
||||
│ │ └─────┘ └──────┘ └───────┘ │ │
|
||||
│ └────────────────────────────────────┘ │
|
||||
│ ▲ │
|
||||
│ │ │
|
||||
└───────────────────┼────────────────────────┘
|
||||
│
|
||||
┌───────────────────┼────────────────────────┐
|
||||
│ SOURCE DIRECTORIES (13 ecosystems) │
|
||||
│ ~/.kimi/ ~/.claude/ ~/.hermes/ │
|
||||
│ ~/.opencode/ ~/.copilot/ ~/.gemini/ │
|
||||
│ ~/.zai/ ~/.zcode/ ~/.mimocode/ ... │
|
||||
└────────────────────────────────────────────┘
|
||||
```
|
||||
|
||||
### Data Flow
|
||||
|
||||
```
|
||||
Agent ecosystems on disk
|
||||
│
|
||||
▼
|
||||
Importers (Python) ← Normalize to NormalizedSession/Message
|
||||
│
|
||||
▼
|
||||
Brain Store (filesystem) ← session-<source>-<id>/session.jsonl
|
||||
│
|
||||
▼
|
||||
GBrain Engine (TypeScript) ← Ingest, index (SQLite FTS5), serve
|
||||
│
|
||||
├─ HTTP API (/api/brain/*) → Theia, VS Code, TUI, CLI
|
||||
└─ MCP Server (stdio) → Hermes Agent
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 4. Tech Stack
|
||||
|
||||
| Layer | Technology | Version |
|
||||
|-------|-----------|---------|
|
||||
| **Importers** | Python | 3.13 |
|
||||
| **GBrain Engine** | TypeScript + Node.js | TS 5.3, Node 20 |
|
||||
| **Database** | SQLite + FTS5 | built-in |
|
||||
| **HTTP Server** | Express | latest |
|
||||
| **CLI Framework** | Commander | latest |
|
||||
| **IDE** | Eclipse Theia | 1.73.1 |
|
||||
| **Frontend** | React | 19 |
|
||||
| **Agent Protocol** | MCP SDK | @modelcontextprotocol/sdk |
|
||||
| **HTTP Client** | Axios | latest |
|
||||
|
||||
---
|
||||
|
||||
## 5. Execution Plan
|
||||
|
||||
### Track A — Agent Importers (Sessions 1–4)
|
||||
|
||||
**Goal:** Cover all 13 source directories with working adapters.
|
||||
|
||||
#### Phase A1: Refactor adapters into a package
|
||||
**Session:** 1 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description | Files |
|
||||
|------|-------------|-------|
|
||||
| 1 | Create `adapters/base.py` with shared helpers (`_load_json`, `_load_jsonl`, `_file_times`, `HOME`) | NEW: `adapters/base.py` |
|
||||
| 2 | Move each existing adapter into its own module | NEW: `adapters/kimi.py`, `hermes.py`, `claude_code.py`, `antigravity.py`, `qwen_code.py`, `mimocode.py` |
|
||||
| 3 | Update `__init__.py` to re-export all `discover_*` functions | MODIFY: `adapters/__init__.py` |
|
||||
| 4 | Regression test: `python3 scripts/sync-agents-to-brain.py --dry-run --summary` — same counts as before | VERIFY |
|
||||
| 5 | Commit | |
|
||||
|
||||
**Exit gate:** Same session counts as monolithic `adapters.py`. No regressions.
|
||||
|
||||
#### Phase A2: Add kimi-code adapter
|
||||
**Session:** 1 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.kimi-code/` structure (likely same format as Kimi) |
|
||||
| 2 | Write `discover_kimi_code()` yielding `NormalizedSession` objects |
|
||||
| 3 | Register in `__init__.py` and dispatch dict |
|
||||
| 4 | Dry-run: `python3 scripts/sync-agents-to-brain.py --source kimi-code --dry-run --summary` |
|
||||
| 5 | Commit |
|
||||
|
||||
#### Phase A3: Add antigravity-ide adapter
|
||||
**Session:** 2 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.antigravity-ide/` (likely JSON/JSONL or SQLite) |
|
||||
| 2 | Write adapter; reuse `sqlite3` pattern from `antigravity.py` if SQLite |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A4: Add gemini adapter
|
||||
**Session:** 2 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.gemini/` — may contain `antigravity-cli/` (covered) + other tools |
|
||||
| 2 | Write adapter for non-antigravity Gemini artifacts |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A5: Add copilot adapter
|
||||
**Session:** 3 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.copilot/` (likely VS Code extension state or CLI cache) |
|
||||
| 2 | Write adapter |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A6: Add opencode adapter
|
||||
**Session:** 3 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.opencode/` (likely JSONL sessions) |
|
||||
| 2 | Write adapter |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A7: Add zai adapter
|
||||
**Session:** 4 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.zai/` |
|
||||
| 2 | Write adapter |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A8: Add zcode adapter
|
||||
**Session:** 4 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inspect `~/.zcode/` |
|
||||
| 2 | Write adapter |
|
||||
| 3 | Register and dry-run |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase A9: Full import run & verification
|
||||
**Session:** 4 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description | Command |
|
||||
|------|-------------|---------|
|
||||
| 1 | Run full import | `python3 scripts/sync-agents-to-brain.py --skip-active --summary` |
|
||||
| 2 | Verify counts per source | `ls .aurelio/brain/ \| grep "^session-" \| sed 's/session-\([^-]*\)-.*/\1/' \| sort \| uniq -c \| sort -rn` |
|
||||
| 3 | Spot-check one session from each new source for correct format | Manual |
|
||||
| 4 | Commit | |
|
||||
|
||||
**🟢 Exit Gate A:** All 13 sources have adapters. Full dry-run completes without errors. ≥8 sources produce sessions (some dirs may be empty on this host).
|
||||
|
||||
---
|
||||
|
||||
### Track B — GBrain Engine (Sessions 5–9)
|
||||
|
||||
**Goal:** Build `aurelio-gbrain` as a real TypeScript engine.
|
||||
|
||||
#### Phase B1: Project scaffold
|
||||
**Session:** 5 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description | Files |
|
||||
|------|-------------|-------|
|
||||
| 1 | `package.json` with deps: axios, better-sqlite3, express, commander, typescript, @types/node | NEW: `aurelio-gbrain/package.json` |
|
||||
| 2 | `tsconfig.json` targeting Node 20, strict mode | NEW: `aurelio-gbrain/tsconfig.json` |
|
||||
| 3 | `types.ts` — `BrainSession`, `BrainMessage`, `SearchResult`, `BrainStats` | NEW: `src/types.ts` |
|
||||
| 4 | `brain.ts` — `GBrain` class with `ingest()`, `search()`, `listSessions()`, `getSession()`, `stats()` | NEW: `src/brain.ts` |
|
||||
| 5 | `cli.ts` — commands: `ingest`, `search`, `list`, `stats`, `serve` | NEW: `src/cli.ts` |
|
||||
| 6 | Build and smoke-test: `npm install && npm run build && node dist/cli.js stats` | VERIFY |
|
||||
| 7 | Commit | |
|
||||
|
||||
#### Phase B2: SQLite FTS5 index
|
||||
**Session:** 6 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description | Files |
|
||||
|------|-------------|-------|
|
||||
| 1 | `indexer.ts` — `BrainIndexer` class with `indexSession()`, `search()`, `rebuild()` | NEW: `src/indexer.ts` |
|
||||
| 2 | Wire `GBrain` to use `BrainIndexer` for search, fallback to filesystem scan if stale | MODIFY: `src/brain.ts` |
|
||||
| 3 | Add `reindex` CLI command | |
|
||||
| 4 | Test: `node dist/cli.js reindex && node dist/cli.js search "universalis"` | VERIFY |
|
||||
| 5 | Commit | |
|
||||
|
||||
#### Phase B3: HTTP API server
|
||||
**Session:** 7 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | `server.ts` — Express app with 6 endpoints |
|
||||
| 2 | Add `serve` CLI command with `--port` flag (default 18000) |
|
||||
| 3 | Test all endpoints with `curl` |
|
||||
| 4 | Commit |
|
||||
|
||||
**API Endpoints:**
|
||||
|
||||
| Method | Path | Returns |
|
||||
|--------|------|---------|
|
||||
| `GET` | `/api/brain/health` | `{ status: 'ok' }` |
|
||||
| `GET` | `/api/brain/stats` | `BrainStats` |
|
||||
| `GET` | `/api/brain/sessions?source=&limit=` | `BrainSession[]` |
|
||||
| `GET` | `/api/brain/sessions/:id` | `BrainSession` |
|
||||
| `GET` | `/api/brain/search?q=&limit=` | `SearchResult[]` |
|
||||
| `POST` | `/api/brain/ingest` | triggers re-ingest |
|
||||
|
||||
#### Phase B4: Dream cycle / synthesis
|
||||
**Session:** 8 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | `dream.ts` — `DreamCycle` class: `run()`, `summarizeSession()`, `extractKnowledge()`, `linkSessions()` |
|
||||
| 2 | Add `dream` CLI command |
|
||||
| 3 | Add `POST /api/brain/dream` endpoint |
|
||||
| 4 | Test with small batch |
|
||||
| 5 | Commit |
|
||||
|
||||
**Note:** MVP dream cycle is extractive (no LLM). LLM-enhanced dream is a stretch goal.
|
||||
|
||||
#### Phase B5: MCP server surface
|
||||
**Session:** 9 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | `mcp.ts` — MCP server exposing 4 tools: `brain_search`, `brain_list_sessions`, `brain_get_session`, `brain_stats` |
|
||||
| 2 | Add `mcp` CLI command (stdio transport) |
|
||||
| 3 | Register in `.aurelio/mcp_config.json` |
|
||||
| 4 | Test with `mcp-remote` or direct stdio |
|
||||
| 5 | Commit |
|
||||
|
||||
**🟢 Exit Gate B:** GBrain builds (`npm run build`), CLI works, HTTP server responds, FTS5 search returns results, MCP server lists tools. `curl http://localhost:18000/api/brain/stats` returns non-zero session count.
|
||||
|
||||
---
|
||||
|
||||
### Track C — Theia Backend Integration (Sessions 10–13)
|
||||
|
||||
**Goal:** Replace the thin scaffold with a real backend that talks to GBrain.
|
||||
|
||||
#### Phase C1: GBrain client package
|
||||
**Session:** 10 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description | Files |
|
||||
|------|-------------|-------|
|
||||
| 1 | `GBrainClient` class with `stats()`, `listSessions()`, `getSession()`, `search()`, `ingest()`, `dream()` | NEW: `aurelio-theia/packages/aurelio-gbrain-client/` |
|
||||
| 2 | Add to Theia workspace `package.json` workspaces | MODIFY |
|
||||
| 3 | Build: `npm run build --workspaces` | VERIFY |
|
||||
| 4 | Commit | |
|
||||
|
||||
#### Phase C2: Real Theia backend service
|
||||
**Session:** 11 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Rewrite `AurelioBackendService` — inject `GBrainClient` via `AURELIO_GBRAIN_ENDPOINT` env var (default `http://192.168.0.38:18000`) |
|
||||
| 2 | Implement real methods: `syncPush()`, `syncPull()`, `getHealth()`, `searchBrain()`, `listSessions()`, `getSession()`, `getStats()`, `triggerDream()` |
|
||||
| 3 | Add Theia RPC endpoints |
|
||||
| 4 | Build and verify |
|
||||
| 5 | Commit |
|
||||
|
||||
#### Phase C3: Wire frontend commands to backend
|
||||
**Session:** 12 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Inject `AurelioBackendService` (via RPC proxy) into `AurelioCommandContribution` |
|
||||
| 2 | Wire high-value commands first: `syncPush`/`syncPull`, `startSession`, `viewSessionHistory`, `showAgentSummary`, `suggestPrompts`, `addContext` |
|
||||
| 3 | Keep low-value commands as stubs (hardware dashboards, SPICE, etc.) |
|
||||
| 4 | Build and verify |
|
||||
| 5 | Commit |
|
||||
|
||||
#### Phase C4: Brain widget
|
||||
**Session:** 13 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | `brain-widget.tsx` — React component: session list (filterable by source), search bar, session detail view, stats panel |
|
||||
| 2 | `brain-widget-contribution.ts` — register with Theia widget manager |
|
||||
| 3 | Wire `aurelio.viewSessionHistory` command to open widget |
|
||||
| 4 | Build and verify |
|
||||
| 5 | Commit |
|
||||
|
||||
**🟢 Exit Gate C:** Theia builds without errors. `AurelioBackendService` responds to health checks. ≥5 frontend commands call real backend methods. Brain widget opens and displays sessions.
|
||||
|
||||
---
|
||||
|
||||
### Track D — Fleet Sync & Surfaces (Sessions 14–16)
|
||||
|
||||
**Goal:** Deploy to CT 208, expose to all surfaces.
|
||||
|
||||
#### Phase D1: Deploy GBrain to CT 208
|
||||
**Session:** 14 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Write systemd unit file for GBrain HTTP server (port 18000) |
|
||||
| 2 | Write deploy script: build → copy to CT 208 → install systemd unit → start |
|
||||
| 3 | Deploy and verify: `curl http://192.168.0.38:18000/api/brain/health` |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase D2: Sync brain to CT 208
|
||||
**Session:** 14 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Verify `sync.py --push` pushes `.aurelio/brain/` |
|
||||
| 2 | Run push and verify CT 208 has sessions |
|
||||
| 3 | Commit |
|
||||
|
||||
#### Phase D3: VS Code extension integration
|
||||
**Session:** 15 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Add `GBrainClient` to VS Code extension |
|
||||
| 2 | Add commands: `Aurelio: Search Brain`, `Aurelio: List Sessions`, `Aurelio: Show Stats` |
|
||||
| 3 | Build VSIX and deploy to CT 205 |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase D4: TUI + CLI integration
|
||||
**Session:** 15 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Add brain search/list commands to TUI |
|
||||
| 2 | Add brain search/list commands to CLI |
|
||||
| 3 | Build and test |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase D5: Hermes Agent integration
|
||||
**Session:** 16 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Register GBrain MCP server in `.aurelio/mcp_config.json` |
|
||||
| 2 | Test from Hermes Agent |
|
||||
| 3 | Verify Hermes can search brain and get results |
|
||||
| 4 | Commit |
|
||||
|
||||
#### Phase D6: End-to-end verification
|
||||
**Session:** 16 · **Status:** 🔴 NOT STARTED
|
||||
|
||||
| Step | Description |
|
||||
|------|-------------|
|
||||
| 1 | Run importer on all sources |
|
||||
| 2 | Push brain to CT 208 |
|
||||
| 3 | Verify GBrain on CT 208 has all sessions |
|
||||
| 4 | Verify Theia IDE can search brain |
|
||||
| 5 | Verify VS Code extension can search brain |
|
||||
| 6 | Verify Hermes Agent can search brain via MCP |
|
||||
| 7 | Document any gaps |
|
||||
|
||||
**🟢 Exit Gate D:** GBrain runs on CT 208 as a systemd service. All 4 surfaces (Theia, VS Code, TUI/CLI, Hermes) can query the brain. Full import → push → query cycle completes without errors.
|
||||
|
||||
---
|
||||
|
||||
## 6. Hard Rules
|
||||
|
||||
1. **Never edit UOS source** (per mycelium directive). This plan is entirely in the Aurélio ecosystem.
|
||||
2. **No second Hermes gateway.** Theia backend talks to GBrain via HTTP, not via Hermes tools.
|
||||
3. **Content-addressed writes only.** Brain session directories named `session-<source>-<id>`, written only when content changes.
|
||||
4. **All new adapters must be defensive.** Missing source directory → yield nothing. Malformed record → skip, never fatal.
|
||||
5. **Idempotency is non-negotiable.** Re-running the importer must produce zero writes when nothing changed.
|
||||
|
||||
---
|
||||
|
||||
## 7. Risks & Mitigations
|
||||
|
||||
| Risk | Impact | Mitigation |
|
||||
|------|--------|-----------|
|
||||
| **Source format variance** — each agent stores sessions differently (SQLite, JSONL, protobuf) | Medium | Adapters are defensive; unknown formats skipped with warning, not fatal |
|
||||
| **Empty source directories** — hermes, claude-code, etc. may have 0 sessions on this host | Low | Importer handles gracefully; user may expect data that isn't there |
|
||||
| **GBrain scope creep** — dream cycle involves LLM calls | Medium | MVP dream cycle is extractive (no LLM); LLM-enhanced dream is stretch goal |
|
||||
| **Theia build complexity** — Theia 1.73.1 + React 19 is bleeding edge | High | Existing scaffold already builds; add incrementally |
|
||||
| **CT 208 deployment** — systemd + Node may have permission issues | Medium | Deploy script handles sudo; fallback: Docker container |
|
||||
|
||||
---
|
||||
|
||||
## 8. Tradeoffs
|
||||
|
||||
| Decision | Chosen | Alternative | Rationale |
|
||||
|----------|--------|-------------|-----------|
|
||||
| **GBrain as HTTP service first** | ✅ Service | Library | Simpler to deploy and debug; HTTP contract stays same when it becomes a library inside Gabinete Hub later |
|
||||
| **SQLite FTS5 over vector DB** | ✅ FTS5 | Embeddings | FTS5 is good enough for keyword search; vector embeddings are a stretch goal |
|
||||
| **Wire 5–10 high-value commands first** | ✅ Subset | All 50+ | Pragmatic; remaining stubs until user prioritizes |
|
||||
| **Adapter per file** | ✅ Split package | Monolithic | Prevents `adapters.py` from becoming a 2000-line monster |
|
||||
|
||||
---
|
||||
|
||||
## 9. Open Questions
|
||||
|
||||
1. **Which agent ecosystems beyond the 13 listed?** Cursor, Windsurf, Aider, Continue, Cody, Tabnine, etc. The adapter pattern makes adding #14+ a 30-minute task — but we need to know which ones you actually use.
|
||||
2. **Should GBrain run on CT 208 or on your workstation?** Current plan: CT 208 primary, local cache secondary. Want local-first with sync?
|
||||
3. **Dream cycle LLM provider?** Which provider from the registry? Default `openrouter` or `kimi-coding`?
|
||||
4. **VS Code extension repo structure?** Is `aurelio-vscode` a submodule of `replica-omnisciente` or standalone? Paths change depending on answer.
|
||||
|
||||
---
|
||||
|
||||
## 10. Success Metrics
|
||||
|
||||
| Metric | Target |
|
||||
|--------|--------|
|
||||
| **Coverage** | 13/13 source directories have adapters |
|
||||
| **Import speed** | Full import completes in <5 minutes |
|
||||
| **Import yield** | Sessions from ≥8 sources (some dirs may be empty) |
|
||||
| **Search latency** | GBrain FTS5 search returns results in <100ms |
|
||||
| **API latency** | HTTP server responds to all 6 endpoints in <200ms |
|
||||
| **IDE integration** | Theia brain widget opens and displays sessions without errors |
|
||||
| **Fleet coverage** | All 4 surfaces can query the brain |
|
||||
| **Idempotency** | Re-running import produces 0 writes when nothing changed |
|
||||
|
||||
---
|
||||
|
||||
## 11. Files Likely to Change
|
||||
|
||||
### Track A
|
||||
- `replica-omnisciente/scripts/agent_importers/adapters/*.py` (new package)
|
||||
- `replica-omnisciente/scripts/agent_importers/__init__.py`
|
||||
- `replica-omnisciente/scripts/sync-agents-to-brain.py`
|
||||
|
||||
### Track B
|
||||
- `replica-omnisciente/aurelio-gbrain/` (entire new package)
|
||||
|
||||
### Track C
|
||||
- `aurelio-theia/AurelioTheia/src/node/aureilio-backend-module.ts`
|
||||
- `aurelio-theia/AurelioTheia/src/browser/aureilio-frontend-module.ts`
|
||||
- `aurelio-theia/AurelioTheia/src/browser/brain-widget.tsx` (new)
|
||||
- `aurelio-theia/packages/aurelio-gbrain-client/` (new package)
|
||||
|
||||
### Track D
|
||||
- `replica-omnisciente/.aurelio/sync.py`
|
||||
- `replica-omnisciente/.aurelio/mcp_config.json`
|
||||
- `aurelio-vscode/src/` (integration)
|
||||
- `aurelio-tui/src/` (integration)
|
||||
- `aurelio-cli/src/` (integration)
|
||||
|
||||
---
|
||||
|
||||
## 12. References
|
||||
|
||||
| Document | Path |
|
||||
|----------|------|
|
||||
| Ecosystem integration pattern | `references/aurelio-ecosystem-integration.md` |
|
||||
| Strategic master plans methodology | `references/strategic-master-plans.md` |
|
||||
| Guest → host import pattern | `references/importing-guest-component-into-host.md` |
|
||||
| Fleet map & build commands | `replica-omnisciente/AGENTS.md` |
|
||||
| Existing importer entry point | `scripts/sync-agents-to-brain.py` |
|
||||
| Normalized session contract | `scripts/agent_importers/engine.py` |
|
||||
| 50+ command stubs | `aurelio-theia/AurelioTheia/src/browser/aurelio-frontend-module.ts` |
|
||||
| Thin backend scaffold | `aurelio-theia/AurelioTheia/src/node/aureilio-backend-module.ts` |
|
||||
|
||||
---
|
||||
|
||||
## 13. Immediate Next Actions
|
||||
|
||||
1. **Confirm the 13 source directories exist** on your host and inspect their formats (`ls` + `file` on each).
|
||||
2. **Start Track A Phase A1:** Refactor adapters into a package.
|
||||
3. **In parallel, start Track B Phase B1:** Scaffold GBrain package.
|
||||
4. **Report back** with findings from source directory inspection — some adapters may be trivial (same format as existing), others may need research.
|
||||
|
||||
---
|
||||
|
||||
## 14. Subagent-Report Log
|
||||
|
||||
| Date | Subagent | Task | Report |
|
||||
|------|----------|------|--------|
|
||||
| 2026-07-12 | Hermes | On-disk grounding | Confirmed: aurelio-theia is thin scaffold, aurelio-gbrain is empty dir, 6/13 adapters exist, brain has 539 kimi + 127 claude + 3 antigravity + 2 mimocode + 1 hermes sessions (~800 total). |
|
||||
| 2026-07-13 | Hermes | Fleet plan audit | All 16 phases across 4 tracks verified as NOT STARTED. 13 source directories mapped. |
|
||||
## Visionary Scope
|
||||
|
||||
**Aurelio OmniBrain** is the omniscient orchestration / execution layer that
|
||||
ties the Portugal Futurista fleet together: it observes every replica and
|
||||
realm, plans across them, and drives execution through the existing
|
||||
toolchains (mycelium agent blocks, Aurelio Theia, UniversalisOS, GBrain).
|
||||
|
||||
It is the "brain" that makes the `replica-omnisciente` realms (e.g.
|
||||
`maquina-na-mao`, `nervura-electrica`, `smart-device-firmware`) act as one
|
||||
coordinated system rather than isolated projects.
|
||||
|
||||
## Architecture Approach
|
||||
|
||||
* **Observation** — ingest state/telemetry from each realm (MQTT/HTTP as used
|
||||
by `nervura-electrica` today).
|
||||
* **Planning** — model cross-realm intent as mycelium agent-block workspaces
|
||||
(IP-XACT-style: agents = components, ports = ACP channels, memory = context).
|
||||
* **Execution** — drive UniversalisOS + Aurelio Theia to instantiate agents and
|
||||
deploy/config via mycelium codegen.
|
||||
* **Memory** — GBrain as the persistent, cross-session state of the brain.
|
||||
|
||||
## Cross-Realm Integration Points
|
||||
|
||||
| Realm / Repo | Role in OmniBrain |
|
||||
| ---------------------- | -------------------------------------------------- |
|
||||
| `replica-omnisciente` | Fleet of replicas; observation + execution targets|
|
||||
| `maquina-na-mao` | Mobile/desktop operator surface (Flutter) |
|
||||
| `nervura-electrica` | Backend (MQTT/HTTP), primary telemetry bus |
|
||||
| `smart-device-firmware`| Edge firmware (Ode Triunfal / Guiador) |
|
||||
| `mycelium` | XSD→codegen; agent-block schema + emit targets |
|
||||
| `aurelio-theia` | IDE; clickable agent blocks; design canvas |
|
||||
| `universalisos` | Target OS / runtime for instantiated agents |
|
||||
| `gbrain` | Persistent memory / reasoning state |
|
||||
|
||||
## Execution Tracker
|
||||
|
||||
Legend: ✅ done · 🟡 in progress · ⏳ queued · 🚫 out of scope
|
||||
|
||||
### Phase 0 — Foundations
|
||||
| Task | Status | Notes |
|
||||
| ---- | ------ | ----- |
|
||||
| Define OmniBrain scope & non-negotiables | ⏳ | confirm this scaffold |
|
||||
| Inventory realm observation surfaces (MQTT topics, HTTP APIs) | ⏳ | start with `nervura-electrica` |
|
||||
| Pin mycelium agent-block schema (`conformance/agentic/uos-agent-blocks-1.0.xsd`) | ⏳ | pull from master plan A.2 |
|
||||
|
||||
### Phase 1 — Observation
|
||||
| Task | Status | Notes |
|
||||
| ---- | ------ | ----- |
|
||||
| Telemetry bus adapter for `nervura-electrica` | ⏳ | reuse `maquina-na-mao` MQTT client pattern |
|
||||
| Per-realm state model (devices, replicas, health) | ⏳ | polymorphic like `OdeTriunfalNode` |
|
||||
| Unified observation feed into GBrain | ⏳ | |
|
||||
|
||||
### Phase 2 — Planning
|
||||
| Task | Status | Notes |
|
||||
| ---- | ------ | ----- |
|
||||
| Agent-block workspace format for cross-realm plans | ⏳ | IP-XACT mapping (master plan A.1) |
|
||||
| Planner → agent-block workspace compiler | ⏳ | mycelium codegen target |
|
||||
| Conflict / resource arbitration across realms | ⏳ | |
|
||||
|
||||
### Phase 3 — Execution
|
||||
| Task | Status | Notes |
|
||||
| ---- | ------ | ----- |
|
||||
| Instantiate agents via Aurelio Theia blocks | ⏳ | |
|
||||
| Deploy/config to UniversalisOS from mycelium emit | ⏳ | |
|
||||
| Closed-loop execution feedback to GBrain | ⏳ | |
|
||||
|
||||
## Open Questions
|
||||
1. Is "OmniBrain" == GBrain, or a separate orchestration layer above it?
|
||||
2. Which realm is the first vertical slice (suggest `nervura-electrica` ↔ `maquina-na-mao`)?
|
||||
3. Target runtime for instantiated agents — UniversalisOS only, or also edge firmware?
|
||||
4. Confirm this tracker replaces/augments the 2026-07-12 master plans.
|
||||
|
|
|
|||
295
.aurelio/plans/2026-07-30_aurelio-consolidation-audit.md
Normal file
295
.aurelio/plans/2026-07-30_aurelio-consolidation-audit.md
Normal file
|
|
@ -0,0 +1,295 @@
|
|||
# Aurelio Consolidation Audit — 30/07/2026
|
||||
|
||||
> Pre-Hetzner migration. Full state analysis of replica-omnisciente + service topology truth.
|
||||
|
||||
---
|
||||
|
||||
## 1. REPO: replica-omnisciente
|
||||
|
||||
**5.7GB on disk | .git 618MB (533MB packs) | 7,368 tracked files**
|
||||
|
||||
### 1.1 Real Code Projects (keep, maintain)
|
||||
|
||||
| Project | Lang | Disk (src) | Purpose |
|
||||
|---------|------|-----------|---------|
|
||||
| `dirac/` | TS | 21M | VS Code coding agent (esbuild) |
|
||||
| `tilth/` | Rust | 768K | Code intelligence MCP (tree-sitter) |
|
||||
| `toon/` | TS | 2.1M | Toon format library (pnpm monorepo) |
|
||||
| `shared/` | TS | 12M | Protobuf shared lib (@aurelio/shared) |
|
||||
| `aurelio-theia/` | TS | 3.6M | IDE monorepo **[submodule → Forgejo]** |
|
||||
| `aurelio-gbrain/` | TS/Py | ~3M | GBrain knowledge brain **[submodule → Forgejo]** |
|
||||
| `mycelium-rd/` | Rust | ~1M | XSD→codegen research |
|
||||
| `telminal/` | Py | 376K | Terminal-in-Telegram bot |
|
||||
| `infra/cdp-bridge/` | TS | 164K | CDP bridge |
|
||||
| `infra/lab-gateway/` | Py | ~1M | Hardware lab gateway (ESP32/PPK2/Icicle) |
|
||||
| `infrastructure/fabric/drift/` | Rust | ~100K | Drift agent (SQLite store, plan parser) |
|
||||
| `dashboards/media-explorer/` | TS | 952K | Media explorer dashboard |
|
||||
|
||||
### 1.2 Git Bloat — BINARY BLOBS TRACKED IN HISTORY (618MB .git)
|
||||
|
||||
These are tracked files bloating the git pack (533MB):
|
||||
|
||||
| Path | Size | What |
|
||||
|------|------|------|
|
||||
| `infra/lab-gateway/MPFS_ICICLE_KIT_2025_07/` | ~100M | 27 PolarFire SoC FPGA job/bitstream files |
|
||||
| `realms/tear-de-silicio/standard-app/models/` | ~80M | FPGA bitstreams + VNNX ML models |
|
||||
| `.aurelio/knowledge/*.pdf` | ~32M | 4 research PDFs (airships, final reports) |
|
||||
| `.aurelio/knowledge/session*/` | ~16M | Session artifacts with screenshots/binary |
|
||||
|
||||
**Fix:** `git filter-repo` to purge binaries → move to LFS or external artifact store. Expected .git reduction: ~250-350MB.
|
||||
|
||||
### 1.3 Local-Only Bloat (gitignored, not in repo but wasting disk)
|
||||
|
||||
| Path | Size | Fix |
|
||||
|------|------|-----|
|
||||
| `tilth/target/` | 3.4G | `cargo clean` — rebuild on demand |
|
||||
| `mycelium-rd/target/` | 345M | `cargo clean` |
|
||||
| `infrastructure/fabric/drift/target/` | 570M | `cargo clean` |
|
||||
| `aurelio-gbrain/src/assets/` | 61M | submodule assets, not tracked (ok) |
|
||||
| `.venv-telminal/` | 53M | gitignored (ok) |
|
||||
|
||||
### 1.4 Brain Sessions — 1,090 dirs, 73MB, uncontrolled growth
|
||||
|
||||
| Source | Count | Status |
|
||||
|--------|-------|--------|
|
||||
| kimi | 601 | Synchronized from ~/.kimi/sessions/ |
|
||||
| opencode | 178 | Synchronized |
|
||||
| claude-code | 129 | Synchronized (+ subagents) |
|
||||
| mimo (mo/mp/mq) | 125 | Synchronized |
|
||||
| antigravity | 13 | Synchronized |
|
||||
| hermes | 12 | Synchronized |
|
||||
| uuid (no prefix) | 25 | Legacy/old format |
|
||||
| pi/mimocode | 7 | Synchronized |
|
||||
|
||||
**268 are untracked** (never committed to git). The brain is growing ~50-100 sessions/week.
|
||||
|
||||
**Fix:** Brain sessions should NOT be in git at all. They're runtime data. Add `.aurelio/brain/session-*/` to .gitignore. The GBrain PostgreSQL instance on CT223 is the real persistent store — file-based sessions are redundant once imported.
|
||||
|
||||
### 1.5 Stale/Dead Documentation
|
||||
|
||||
| File | Dated | Problem |
|
||||
|------|-------|---------|
|
||||
| `DIRECTORY_GUIDE.md` | 2026-04-16 | Describes root-level `memory/`, `identity/`, `swarm/` — all moved to `.aurelio/`. Lists wrong realms. |
|
||||
| `AGENT_INDEX.md` | 2025-07-07 | Says CT208 is hub (stopped). 63 projects listed, half don't exist. Wrong IPs. |
|
||||
| `README.md` | 2026-07-06 | Generic, doesn't match actual structure |
|
||||
| `DESIGN.md` | 2026-07-06 | 648 bytes — likely empty stub |
|
||||
| `SETUP_API_KEYS.md` | 2026-07-06 | Probably stale API key instructions |
|
||||
| `task.md` | 2026-07-06 | Stale task file |
|
||||
| `vscode_extension_migration_guide.md` | 2026-07-06 | One-off migration doc, no longer needed |
|
||||
| `IMPLEMENTATION_SUMMARY.md` | 2026-07-11 | Portal consolidation notes |
|
||||
| `PORTAL_CONSOLIDATION_SUMMARY.md` | 2026-07-11 | Same topic |
|
||||
| `NEXT_MONTH_ROADMAP.md` | 2026-07-11 | August 2026 roadmap — expired |
|
||||
|
||||
### 1.6 Empty/Stub Realms (23 dirs, most 8KB empty)
|
||||
|
||||
**Real content (3):** `nervura-electrica` (672K), `universalisos` (308K), `silent-shared-office` (32K)
|
||||
|
||||
**Empty stubs (20):** aprendiz-de-sensacoes, matriz-da-maquina, ode-electromagnetica, orquestra-de-hardware, alquimista, etc.
|
||||
|
||||
**Fix:** Delete empty stubs. Move realms/ with real content to top-level project docs or archive.
|
||||
|
||||
---
|
||||
|
||||
## 2. SERVICE TOPOLOGY — TRUTH vs DOCUMENTATION
|
||||
|
||||
### 2.1 What the 07-16 Audit Says (WRONG)
|
||||
|
||||
```
|
||||
CT208 (192.168.0.125) = ALL Aurelio services
|
||||
gbrain :18001, gabinete :18000, bot-rs, 7 MCP servers, 6 kimi swarm agents
|
||||
ASUS = fabricd :8081, orchestrator :8200
|
||||
```
|
||||
|
||||
### 2.2 What Memory Says (CURRENT REALITY)
|
||||
|
||||
```
|
||||
CT208 = STOPPED (old hub, decommissioned)
|
||||
CT223 (Gigabyte, dir-rootfs /mnt/disks/usb-sdf) = NEW HUB
|
||||
├── Gabinete :18000
|
||||
├── Bot-rs (Telegram @Aurélio Omnisciente)
|
||||
├── GBrain :18001
|
||||
├── fabricd :8081
|
||||
└── Engineering MCPs
|
||||
CT217 = Hermes gateway 24/7 (Signal/Telegram/Matrix, E2EE)
|
||||
CT235 = Orchestrator :8200
|
||||
CT237 (Gigabyte) = Forgejo :3001 (code.portugalfuturista.org), Woodpecker :8001
|
||||
```
|
||||
|
||||
### 2.3 MCP Config — 16 configured servers, all `npx` (wrong for production)
|
||||
|
||||
`.aurelio/mcp_config.json` lists 16 MCP servers using `npx` commands. But AGENTS.md says they're "remote (mcp-remote to *.portugalfuturista.org)". The config doesn't match reality — these are either:
|
||||
- Run as systemd services on CT223 (sourcing :8081, eda :8082, knowledge :8083, codebase-memory :8084)
|
||||
- Or dead/never-deployed (civil, artist, mechanic, StitchMCP, dart-mcp-server)
|
||||
|
||||
### 2.4 Local MCP servers on disk (.aurelio/mcp/)
|
||||
|
||||
| Server | Size | Status |
|
||||
|--------|------|--------|
|
||||
| `savearth-mcp/` | 84K | Python — active (Savearth fleet) |
|
||||
| `savearth-workspace/` | 60K | Python — active |
|
||||
| `electrical-mcp/` | 720K | TS — active |
|
||||
| `engineering-mcp/` | 340K | TS — active |
|
||||
| `flint-chart-mcp/` | 156K | TS — purpose unclear |
|
||||
| `image-pipe-mcp/` | 100K | TS — purpose unclear |
|
||||
| `aurelio-tools-mcp/` | 92K | TS — purpose unclear |
|
||||
|
||||
### 2.5 Infrastructure Topology (3 PVE hosts)
|
||||
|
||||
```
|
||||
PROXMOX CLUSTER "portugalfuturi"
|
||||
│
|
||||
├── ASUS (192.168.0.38) — 31Gi, PVE 9.2.2
|
||||
│ ├── CT215: pf-matrix (Matrix Synapse :matrix.portugalfuturista.org)
|
||||
│ ├── CT208: OLD HUB — STOPPED
|
||||
│ ├── Cloudflare tunnel pf-tunnel-v2
|
||||
│ └── Other CTs (202-220)
|
||||
│
|
||||
├── GIGABYTE (192.168.0.104) — 46GB, 2x RTX 3070
|
||||
│ ├── CT223: NEW HUB (Gabinete, Bot-rs, GBrain, fabricd, MCPs)
|
||||
│ ├── CT237: Forgejo :3001 + Woodpecker :8001
|
||||
│ ├── CT216: multi-scrobbler (Subsonic)
|
||||
│ ├── Docker media stack (Jellyfin, Sonarr, Radarr, etc.)
|
||||
│ ├── signal-api container :18088
|
||||
│ └── Monitoring (Prometheus + Grafana + Loki + Blackbox)
|
||||
│
|
||||
└── LATTEPANDA (192.168.0.40)
|
||||
├── MinIO :9000/:9001 (S3 / Terraform state)
|
||||
├── LocalStack v3 :4566 (AWS dev)
|
||||
└── Azurite :10000-10002 (Azure dev)
|
||||
|
||||
OFF-LAN:
|
||||
├── CT217 (pf-hermes): Hermes gateway 24/7
|
||||
├── CT235: Orchestrator :8200
|
||||
├── VPS pf-edge (94.130.24.119): WG endpoint
|
||||
└── CTs 213/231/234/235/236: Savearth fleet
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 3. HETZNER MIGRATION — TARGET ARCHITECTURE
|
||||
|
||||
### 3.1 What Should Move to Hetzner (cloud-facing services)
|
||||
|
||||
These services benefit from public IP, low latency to external APIs, and 24/7 uptime without LAN dependency:
|
||||
|
||||
| Service | Current | Hetzner Target | Why |
|
||||
|---------|---------|----------------|-----|
|
||||
| Hermes gateway | CT217 | Hetzner VM | 24/7 messaging, no LAN dependency |
|
||||
| GBrain (PostgreSQL) | CT223 | Hetzner VM (with pg backup) | Central knowledge store, API-facing |
|
||||
| Gabinete API | CT223 | Hetzner VM | Central API, bot backend |
|
||||
| Bot-rs (Telegram) | CT223 | Hetzner VM | Telegram webhook needs public IP |
|
||||
| Forgejo + CI | CT237 | Hetzner VM | Git push/pull latency, public access |
|
||||
| Cloudflare tunnel endpoint | ASUS | Cloudflare→Hetzner | Remove single LAN point of failure |
|
||||
|
||||
### 3.2 What Should Stay on LAN (hardware-dependent, latency-sensitive)
|
||||
|
||||
| Service | Host | Why |
|
||||
|---------|------|-----|
|
||||
| Hardware lab gateway | ASUS/CT206 | Physical ESP32/PPK2/Icicle connected |
|
||||
| Media stack | Gigabyte | Local storage, GPU transcoding |
|
||||
| Signal-API | Gigabyte | Phone registration tied to LAN |
|
||||
| Cloud emulators | Lattepanda | Dev-only, LAN-local |
|
||||
| Savearth fleet CTs | Various | Hardware-attached (ESP32/PPK2) |
|
||||
| Proxmox cluster itself | All 3 hosts | Bare metal |
|
||||
|
||||
### 3.3 Service Separation Principle
|
||||
|
||||
Current problem: CT223 is a god-container running everything. Hetzner migration should split:
|
||||
|
||||
```
|
||||
HETZNER (public cloud)
|
||||
├── VM-1: aurelio-core (Gabinete + Bot-rs + GBrain)
|
||||
├── VM-2: aurelio-forge (Forgejo + Woodpecker CI)
|
||||
├── VM-3: aurelio-gateway (Hermes + Cloudflare tunnel endpoint)
|
||||
└── VM-4: aurelio-monitoring (Grafana/Prometheus mirror)
|
||||
|
||||
LAN (hardware-locked)
|
||||
├── ASUS: lab hardware, local CTs
|
||||
├── Gigabyte: media, signal, CT223→demoted
|
||||
├── Lattepanda: dev emulators
|
||||
└── Savearth fleet: hardware testing
|
||||
|
||||
VPN: WireGuard mesh Hetzner ↔ LAN (existing pf-edge WG)
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 4. REPO CLEANUP ACTIONS
|
||||
|
||||
### 4.1 Immediate (low risk)
|
||||
|
||||
1. **`cargo clean` all Rust projects** — frees 4.3GB local disk
|
||||
```
|
||||
cd tilth && cargo clean
|
||||
cd mycelium-rd && cargo clean
|
||||
cd infrastructure/fabric/drift && cargo clean
|
||||
```
|
||||
|
||||
2. **Delete empty realm stubs** (20 dirs × 8KB)
|
||||
```
|
||||
realms/aprendiz-de-sensacoes, matriz-da-maquina, ode-electromagnetica, ...
|
||||
```
|
||||
|
||||
3. **Delete stale root docs**
|
||||
```
|
||||
DESIGN.md, task.md, SETUP_API_KEYS.md, vscode_extension_migration_guide.md,
|
||||
IMPLEMENTATION_SUMMARY.md, PORTAL_CONSOLIDATION_SUMMARY.md, NEXT_MONTH_ROADMAP.md
|
||||
```
|
||||
|
||||
4. **Add brain sessions to .gitignore**
|
||||
```
|
||||
.aurelio/brain/session-*/
|
||||
```
|
||||
|
||||
### 4.2 Medium (requires git history rewrite)
|
||||
|
||||
5. **Purge binary blobs from git** — saves ~250MB from .git
|
||||
```
|
||||
git filter-repo --path infra/lab-gateway/MPFS_ICICLE_KIT_2025_07 --invert-paths
|
||||
git filter-repo --path realms/tear-de-silicio/standard-app/models --invert-paths
|
||||
git filter-repo --path-glob '.aurelio/knowledge/*.pdf' --invert-paths
|
||||
```
|
||||
Move to: MinIO bucket or Git LFS (if Forgejo LFS is enabled — currently NOT per AGENTS.md)
|
||||
|
||||
### 4.3 Documentation Rewrite
|
||||
|
||||
6. **Rewrite `DIRECTORY_GUIDE.md`** to reflect `.aurelio/` structure (not root-level)
|
||||
7. **Rewrite `AGENT_INDEX.md`** with real CT topology (CT223 hub, not CT208)
|
||||
8. **Update `AGENTS.md`** infra targets table to match memory reality
|
||||
9. **Delete or archive** the 07-16 audit (superseded by this document)
|
||||
|
||||
---
|
||||
|
||||
## 5. SERVICE CONSOLIDATION ACTIONS
|
||||
|
||||
### 5.1 MCP Server Cleanup
|
||||
|
||||
16 MCP servers configured but reality is unclear. Need to:
|
||||
1. SSH to CT223, enumerate actual systemd services
|
||||
2. Compare against `.aurelio/mcp_config.json`
|
||||
3. Remove dead MCP entries (civil, artist, mechanic, StitchMCP, dart-mcp-server)
|
||||
4. Consolidate overlapping: `electrical-mcp` vs `engineering-mcp` vs `electrical-eda-mcp` vs `electrical-sourcing-mcp` — 4 servers for electrical?
|
||||
|
||||
### 5.2 Kimi Swarm Decommission?
|
||||
|
||||
6 Kimi swarm agents on old CT208 (orchestrator, log-processor, ota-validator, regression-detector, report-synthesizer, dashboard). CT208 is stopped. Are these:
|
||||
- Still needed?
|
||||
- Migrated to CT223?
|
||||
- Dead and should be removed from repo (`.aurelio/swarm/`)?
|
||||
|
||||
### 5.3 Lifestream Pipeline Verification
|
||||
|
||||
```
|
||||
Saved Messages → Lifestream userbot → SQLite → Cron 03:00 → GBrain
|
||||
```
|
||||
This pipeline depends on CT223. Verify it still works after migration.
|
||||
|
||||
---
|
||||
|
||||
## 6. OPEN QUESTIONS (need your input)
|
||||
|
||||
1. **Hetzner specs** — What VM(s) did you provision? CPU/RAM/disk? Dedicated or shared?
|
||||
2. **CT223 future** — After Hetzner migration, does CT223 stay as LAN fallback or get repurposed?
|
||||
3. **Git remote strategy** — Keep GitHub (origin) + Forgejo? Or consolidate to Forgejo-only now that Hetzner gives public access?
|
||||
4. **Brain persistence** — GBrain PostgreSQL on Hetzner means migrating 1362+ pages. Acceptable? Or keep DB on LAN with Hetzner as read-replica?
|
||||
5. **Kimi swarm** — Dead or alive?
|
||||
|
|
@ -0,0 +1,346 @@
|
|||
# Aurelio Consolidation — Execution Plan
|
||||
|
||||
> 30/07/2026. Decisions locked. This is the execution document.
|
||||
|
||||
## DECISIONS (locked by Maestro)
|
||||
|
||||
| Decision | Resolution |
|
||||
|----------|-----------|
|
||||
| Hetzner specs | Agent decides — see sizing below |
|
||||
| CT223 future | Stays as LAN fallback, ready to receive snapshots |
|
||||
| Git remote | Forgejo = primary, GitHub = mirror |
|
||||
| GBrain DB | Stays local (MVP phase), not on Hetzner |
|
||||
| Kimi swarm | Remove from CT223, restructure as reusable sub-agent harness |
|
||||
| Aurelio identity | Aurelio is a **harness** — sub-agents feed into bot, Theia, VS Code |
|
||||
|
||||
## BLOCKER: Hetzner API Token
|
||||
|
||||
The token is in `nervura-electrica/infrastructure/proxmox/iac/tofu/secrets.enc.yaml`,
|
||||
encrypted with age key `age1m56h5m2hra70uppj5jn8w7prynjw8gtjfxpp0tucvgydnnu3u9esaplg9g`.
|
||||
The age private key is NOT on: laptop, pf-edge, ASUS, or CT223.
|
||||
|
||||
**To provision new Hetzner VMs, we need either:**
|
||||
1. The age private key (where was it created?)
|
||||
2. OR a fresh Hetzner API token from console.hetzner.cloud
|
||||
|
||||
---
|
||||
|
||||
## TARGET ARCHITECTURE
|
||||
|
||||
### Sizing Decision
|
||||
|
||||
CT223 currently runs 30+ services on 2GB RAM — it's at the breaking point.
|
||||
Only the compute-intensive AI layer goes to Hetzner. Git forge and gateway
|
||||
stay on LAN (Dell/Gigabyte) for data sovereignty, but are IaC-packaged so
|
||||
they can deploy to any cloud in one command.
|
||||
|
||||
**LAN resources:**
|
||||
- Dell .41: 31Gi RAM, 47GB disk free — 6 CTs running, ample headroom
|
||||
- Gigabyte .104: 46Gi RAM, 86% disk (15GB free) — tight, CT223 is the hog
|
||||
|
||||
**New VMs / CTs:**
|
||||
|
||||
| VM/CT | Host | Type | Specs | Monthly | Purpose |
|
||||
|-------|------|------|-------|---------|---------|
|
||||
| pf-aurelio | **Hetzner** | CPX31 | 2 ded AMD vCPU, 4GB, 160GB | €8.50 | Gabinete, Bot-rs, fabricd, MCP gateway, sub-agent harness |
|
||||
| pf-forge | **Dell .41** | LXC 238 | 4 vCPU, 8GB, 80GB | €0 | Forgejo + Woodpecker CI + GitHub mirror |
|
||||
| pf-stage | **Dell .41** | LXC 239 | 2 vCPU, 4GB, 40GB | €0 | Hermes gateway 24/7 |
|
||||
|
||||
**Total cloud cost: €8.50/month** (only pf-aurelio is cloud).
|
||||
pf-edge (existing CX22) stays as media edge — €4.50/month.
|
||||
Grand total Hetzner: €13/month.
|
||||
|
||||
GBrain stays on CT223 (local, per decision). pf-aurelio connects to it via WireGuard.
|
||||
pf-forge and pf-stage are LAN-local (no WG needed for GBrain access).
|
||||
|
||||
### Service Separation Target
|
||||
|
||||
```
|
||||
HETZNER (public cloud — €13/mo total)
|
||||
│
|
||||
├── pf-edge (94.130.24.119, CX22 — EXISTS)
|
||||
│ ├── Caddy media reverse proxy → LAN via WG
|
||||
│ └── WireGuard → 10.99.0.1 (ASUS)
|
||||
│
|
||||
├── pf-aurelio (NEW, CPX31)
|
||||
│ ├── aurelio-gabinete-rs :18000 → API central
|
||||
│ ├── aurelio-bot-rs → Telegram bot
|
||||
│ ├── aurelio-fabricd :8081 → model gateway
|
||||
│ ├── aurelio-engineering-gw :8090 → federated MCP
|
||||
│ ├── aurelio-subagent-harness → Kimi swarm redesign
|
||||
│ ├── aurelio-web → Olhos-de-Orpheu
|
||||
│ └── brain-sync (→ CT223 GBrain via WG 10.99.0.1)
|
||||
│
|
||||
LAN (Proxmox cluster — data sovereignty, zero cloud cost)
|
||||
│
|
||||
├── DELL .41 (PVE 9.2.2, 31Gi RAM, 47GB free)
|
||||
│ ├── CT213 pf-immich
|
||||
│ ├── CT231 pf-homeassistant
|
||||
│ ├── CT234 pf-marketplace
|
||||
│ ├── CT235 pf-orchestrator :8200
|
||||
│ ├── CT236 pf-savearth
|
||||
│ ├── CT237 pf-iot-stack
|
||||
│ ├── CT238 pf-forge (NEW) — Forgejo :3001 + Woodpecker :8001
|
||||
│ │ ↑ IaC-packaged: swap LXC→cloud in one tofu apply
|
||||
│ └── CT239 pf-stage (NEW) — Hermes gateway 24/7
|
||||
│ ↑ IaC-packaged: swap LXC→cloud in one tofu apply
|
||||
│
|
||||
├── GIGABYTE .104 (46Gi RAM, 86% disk)
|
||||
│ ├── CT223 pf-forja-do-conhecimento — DATA LAYER
|
||||
│ │ ├── pf-gbrain :18001 → STAYS (MVP, local DB)
|
||||
│ │ ├── gbrain-rest-proxy :18002 → STAYS
|
||||
│ │ ├── PostgreSQL (Docker) :5433 → STAYS (GBrain backend)
|
||||
│ │ ├── Docker services → STAYS (Plane, Outline, MinIO)
|
||||
│ │ ├── lifestream-userbot → STAYS (Telethon session)
|
||||
│ │ ├── pf-identity :9091 → STAYS (CA)
|
||||
│ │ └── [snapshot target for pf-aurelio]
|
||||
│ ├── CT237 (old Forgejo — migrate to CT238 Dell, then retire)
|
||||
│ └── Other media CTs (200-222)
|
||||
│
|
||||
├── ASUS .38 (31Gi, PVE cluster primary)
|
||||
│ ├── CT215 pf-matrix (Matrix Synapse)
|
||||
│ ├── CT208 OLD HUB — STOPPED
|
||||
│ ├── Cloudflare tunnel pf-tunnel-v2
|
||||
│ └── WireGuard server 10.99.0.1
|
||||
│
|
||||
WIREGUARD MESH (ASUS = hub, 10.99.0.1):
|
||||
pf-edge 10.99.0.2 ↔ ASUS 10.99.0.1 (media proxy)
|
||||
pf-aurelio 10.99.0.3 ↔ ASUS 10.99.0.1 (→ CT223 GBrain)
|
||||
Dell CTs LAN-direct (no WG needed)
|
||||
```
|
||||
|
||||
### What Moves Off CT223
|
||||
|
||||
| Service | From | To | Route |
|
||||
|---------|------|----|-------|
|
||||
| aurelio-gabinete-rs | CT223 :18000 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-bot-rs | CT223 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-fabricd | CT223 :8081 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-engineering-gw | CT223 :8090 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-native-engine-gw | CT223 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-web | CT223 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| aurelio-webhook | CT223 | pf-aurelio (Hetzner) | direct on VM |
|
||||
| Hermes gateway | CT217 | CT239 Dell (pf-stage) | LAN-direct |
|
||||
| Forgejo + Woodpecker | CT237 Gigabyte | CT238 Dell (pf-forge) | LAN-direct |
|
||||
|
||||
### What Stays on CT223
|
||||
|
||||
| Service | Why it stays |
|
||||
|---------|-------------|
|
||||
| pf-gbrain :18001 | GBrain DB stays local (Maestro decision) |
|
||||
| gbrain-rest-proxy :18002 | Pairs with local GBrain |
|
||||
| PostgreSQL Docker :5433 | GBrain backend DB |
|
||||
| Plane/Outline/MinIO/Redis | Docker stack, heavy data |
|
||||
| lifestream-userbot | Telethon session, LAN-dependent |
|
||||
| pf-identity :9091 | CA + identity management |
|
||||
| brain-sync :8095 | Syncs to local GBrain |
|
||||
| All `pf-*-scribe-208` services | Legacy naming, lightweight |
|
||||
|
||||
### CT223 After Migration
|
||||
|
||||
CT223 drops from 30+ services to ~12. RAM goes from 976Mi used → ~400Mi.
|
||||
It becomes the data layer + fallback. pf-aurelio can snapshot its config
|
||||
to CT223 for disaster recovery.
|
||||
|
||||
---
|
||||
|
||||
## IaC PORTABILITY — LXC TODAY, CLOUD TOMORROW
|
||||
|
||||
pf-forge and pf-stage deploy as Proxmox LXC containers on Dell today.
|
||||
The OpenTofu definitions use a provider-agnostic pattern: swap one
|
||||
variable and the same service deploys to Hetzner/AWS/bare-metal.
|
||||
|
||||
### Structure (extends existing nervura-electrica IaC)
|
||||
|
||||
```
|
||||
nervura-electrica/infrastructure/proxmox/iac/tofu/
|
||||
├── edge_node.tf ← pf-edge (Hetzner, exists)
|
||||
├── aurelio_node.tf ← pf-aurelio (Hetzner CPX31, new)
|
||||
├── forge_node.tf ← pf-forge (Dell LXC 238, new)
|
||||
├── stage_node.tf ← pf-stage (Dell LXC 239, new)
|
||||
├── providers.tf
|
||||
└── secrets.enc.yaml
|
||||
```
|
||||
|
||||
Each node file follows this pattern:
|
||||
|
||||
```hcl
|
||||
# forge_node.tf
|
||||
# Deploys as LXC on Dell today. Set deploy_target="hcloud" to move to cloud.
|
||||
|
||||
variable "forge_deploy_target" {
|
||||
description = "LXC (LAN) or hcloud (Hetzner)"
|
||||
type = string
|
||||
default = "lxc" # ← change this one line to go cloud
|
||||
}
|
||||
|
||||
# LXC variant (Dell .41, zero cost)
|
||||
resource "proxmox_lxc" "forge" {
|
||||
count = var.forge_deploy_target == "lxc" ? 1 : 0
|
||||
vmid = 238
|
||||
target_node = "dell"
|
||||
...
|
||||
}
|
||||
|
||||
# Cloud variant (Hetzner, if needed later)
|
||||
resource "hcloud_server" "forge" {
|
||||
count = var.forge_deploy_target == "hcloud" ? 1 : 0
|
||||
name = "pf-forge"
|
||||
server_type = "cx22"
|
||||
...
|
||||
}
|
||||
```
|
||||
|
||||
Ansible roles handle the service config regardless of where the container/VM runs —
|
||||
same Forgejo config, same Hermes config, same WireGuard mesh entry.
|
||||
|
||||
---
|
||||
|
||||
## KIMI SWARM → AURELIO SUB-AGENT HARNESS
|
||||
|
||||
### Current State (CT223)
|
||||
|
||||
6 Kimi swarm services, all named `pf-*-208.service`:
|
||||
- orchestrator, log-processor, ota-validator, regression-detector, report-synthesizer, dashboard
|
||||
|
||||
These are Python services in `.aurelio/swarm/` that dispatch Kimi API calls
|
||||
as specialized agents. They're alive but isolated — only used by the old
|
||||
swarm dashboard.
|
||||
|
||||
### Redesign: Aurelio Sub-Agent Harness
|
||||
|
||||
Aurelio is a **harness**. The swarm agents become a reusable sub-agent layer
|
||||
that any Aurelio surface can invoke:
|
||||
|
||||
```
|
||||
AURELIO SURFACES (callers)
|
||||
├── Bot-rs (Telegram) → "analyze these logs" → dispatches sub-agent
|
||||
├── Aurelio-Theia (IDE) → user requests review → dispatches sub-agent
|
||||
├── Aurelio VS Code → developer triggers analysis → dispatches sub-agent
|
||||
└── Gabinete API → programmatic dispatch
|
||||
|
||||
SUB-AGENT HARNESS (the swarm, restructured)
|
||||
├── log-processor → parse/stream device/firmware logs
|
||||
├── ota-validator → validate firmware OTA packages
|
||||
├── regression-detector → diff CI runs, flag regressions
|
||||
├── report-synthesizer → aggregate findings into reports
|
||||
└── orchestrator → route requests to the right agent
|
||||
|
||||
TRANSPORT: each surface calls the orchestrator via HTTP/gRPC
|
||||
AUTH: fabricd gateway token (existing)
|
||||
```
|
||||
|
||||
### Implementation
|
||||
|
||||
1. Strip the `pf-*-208` naming and systemd units from CT223
|
||||
2. Package the swarm as a single deployable service (Rust or Python)
|
||||
3. Expose via fabricd as an MCP tool: `aurelio.subagent.dispatch`
|
||||
4. Any surface (bot, IDE, VS Code) calls it through the existing MCP gateway
|
||||
5. Deploy on pf-aurelio alongside Gabinete/Bot-rs
|
||||
|
||||
---
|
||||
|
||||
## REPO CLEANUP (executable now, no Hetzner dependency)
|
||||
|
||||
### Phase 1: Immediate disk recovery
|
||||
|
||||
```bash
|
||||
# Free 4.3GB of local Rust build artifacts
|
||||
cd tilth && cargo clean
|
||||
cd mycelium-rd && cargo clean
|
||||
cd infrastructure/fabric/drift && cargo clean
|
||||
```
|
||||
|
||||
### Phase 2: Git hygiene
|
||||
|
||||
```bash
|
||||
# Stop tracking brain sessions (runtime data, not source)
|
||||
echo '.aurelio/brain/session-*/' >> .gitignore
|
||||
|
||||
# Remove binary blobs from working tree
|
||||
git rm -r --cached infra/lab-gateway/MPFS_ICICLE_KIT_2025_07/
|
||||
git rm -r --cached realms/tear-de-silicio/standard-app/models/
|
||||
```
|
||||
|
||||
### Phase 3: Purge git history (requires force-push)
|
||||
|
||||
```bash
|
||||
# After Phase 2, purge from history (saves ~250MB from .git)
|
||||
git filter-repo --path infra/lab-gateway/MPFS_ICICLE_KIT_2025_07 --invert-paths
|
||||
git filter-repo --path realms/tear-de-silicio/standard-app/models --invert-paths
|
||||
git filter-repo --path-glob '.aurelio/knowledge/*.pdf' --invert-paths
|
||||
|
||||
# Force push (coordinate with all checkouts)
|
||||
git push forgejo main --force
|
||||
git push origin main --force
|
||||
```
|
||||
|
||||
### Phase 4: Delete dead content
|
||||
|
||||
```bash
|
||||
# 20 empty realm stubs
|
||||
rm -rf realms/aprendiz-de-sensacoes realms/matriz-da-maquina ...
|
||||
|
||||
# Stale root docs
|
||||
rm DESIGN.md task.md SETUP_API_KEYS.md vscode_extension_migration_guide.md
|
||||
rm IMPLEMENTATION_SUMMARY.md PORTAL_CONSOLIDATION_SUMMARY.md NEXT_MONTH_ROADMAP.md
|
||||
rm AGENT_INDEX.md DIRECTORY_GUIDE.md
|
||||
```
|
||||
|
||||
### Phase 5: Documentation rewrite
|
||||
|
||||
Rewrite these to match reality:
|
||||
- `AGENTS.md` — update infra targets table (CT223 hub, not CT208)
|
||||
- `README.md` — current repo overview
|
||||
- New `DIRECTORY_GUIDE.md` — reflect `.aurelio/` structure
|
||||
|
||||
---
|
||||
|
||||
## GIT REMOTE STRATEGY
|
||||
|
||||
### Current: dual primary
|
||||
- `origin` = GitHub (fabiorafaelcoutada/replica-omnisciente)
|
||||
- `forgejo` = Forgejo LAN (code.portugalfuturista.org)
|
||||
|
||||
### Target: Forgejo primary, GitHub mirror
|
||||
|
||||
```bash
|
||||
# Swap remotes
|
||||
git remote rename origin github-mirror
|
||||
git remote rename forgejo origin
|
||||
|
||||
# Set up mirror push (Forgejo → GitHub)
|
||||
git remote set-url --add origin --push git@github.com:fabiorafaelcoutada/replica-omnisciente.git
|
||||
```
|
||||
|
||||
Or configure Forgejo's push mirror feature (Forgejo pushes to GitHub automatically on each push).
|
||||
|
||||
---
|
||||
|
||||
## EXECUTION SEQUENCE
|
||||
|
||||
### Done (this session)
|
||||
1. ✅ Full repo audit (5.7GB, 33 CTs, service topology, documentation drift)
|
||||
2. ✅ Unified fleet inventory (`infrastructure/fleet/fleet.yaml`)
|
||||
3. ✅ pfctl CLI — manages ALL nodes (Proxmox + Hetzner) with one command
|
||||
4. ✅ Ansible inventory generator
|
||||
5. ✅ Prometheus scrape config (all nodes)
|
||||
|
||||
### What I can do NOW (no token needed, no destructive ops)
|
||||
6. ✅ Repo cleanup Phase 1-2 (cargo clean, .gitignore)
|
||||
7. ✅ Delete dead realms + stale docs
|
||||
8. ✅ Rewrite AGENTS.md, DIRECTORY_GUIDE.md
|
||||
9. ✅ Write OpenTofu for pf-aurelio (Hetzner, ready when token arrives)
|
||||
10. ✅ Write OpenTofu for pf-forge CT238 + pf-stage CT239 (Dell LXC, deployable now)
|
||||
11. ✅ Write Ansible roles for Forgejo + Hermes (provider-agnostic)
|
||||
|
||||
### What needs the Hetzner token (only pf-aurelio)
|
||||
12. ⏳ Provision pf-aurelio CPX31 on Hetzner
|
||||
13. ⏳ Configure WireGuard pf-aurelio ↔ ASUS
|
||||
14. ⏳ Migrate Gabinete/Bot-rs/fabricd/MCP services to pf-aurelio
|
||||
15. ⏳ Cutover mcp.portugalfuturista.org → pf-aurelio
|
||||
|
||||
### What needs Maestro confirmation (destructive)
|
||||
16. Git history rewrite (force push) — coordinate timing
|
||||
17. Kimi swarm systemd unit removal from CT223 — service interruption window
|
||||
18. Forgejo data migration CT237→CT238 (git repos + DB dump/restore)
|
||||
6
.aurelio/providers/dist/dirac.providers.json
vendored
6
.aurelio/providers/dist/dirac.providers.json
vendored
|
|
@ -44,9 +44,13 @@
|
|||
"value": "anthropic",
|
||||
"label": "Anthropic"
|
||||
},
|
||||
{
|
||||
"value": "google-cloud",
|
||||
"label": "Google Cloud (Vertex AI)"
|
||||
},
|
||||
{
|
||||
"value": "gemini",
|
||||
"label": "Google Gemini"
|
||||
"label": "Google Gemini (AI Studio)"
|
||||
},
|
||||
{
|
||||
"value": "moonshot",
|
||||
|
|
|
|||
17
.aurelio/providers/dist/mcp.providers.json
vendored
17
.aurelio/providers/dist/mcp.providers.json
vendored
|
|
@ -3,6 +3,7 @@
|
|||
"chatFallback": [
|
||||
"kimi-coding",
|
||||
"qwen-code",
|
||||
"google-cloud",
|
||||
"gemini",
|
||||
"anthropic",
|
||||
"openai",
|
||||
|
|
@ -11,6 +12,7 @@
|
|||
"openrouter"
|
||||
],
|
||||
"embedFallback": [
|
||||
"google-cloud",
|
||||
"gemini",
|
||||
"openai",
|
||||
"qwen",
|
||||
|
|
@ -160,9 +162,22 @@
|
|||
"streaming": true,
|
||||
"fallback": true
|
||||
},
|
||||
{
|
||||
"id": "google-cloud",
|
||||
"label": "Google Cloud (Vertex AI)",
|
||||
"auth": "api_key",
|
||||
"keyEnv": "GOOGLE_CLOUD_API_KEY",
|
||||
"baseUrl": "https://us-central1-aiplatform.googleapis.com/v1",
|
||||
"protocol": "gemini",
|
||||
"chatModel": "gemini-2.5-pro",
|
||||
"embeddingModel": "text-embedding-004",
|
||||
"maxContext": 1000000,
|
||||
"streaming": true,
|
||||
"fallback": true
|
||||
},
|
||||
{
|
||||
"id": "gemini",
|
||||
"label": "Google Gemini",
|
||||
"label": "Google Gemini (AI Studio)",
|
||||
"auth": "api_key",
|
||||
"keyEnv": "GEMINI_API_KEY",
|
||||
"baseUrl": "https://generativelanguage.googleapis.com/v1beta",
|
||||
|
|
|
|||
18
.aurelio/providers/dist/portal.providers.json
vendored
18
.aurelio/providers/dist/portal.providers.json
vendored
|
|
@ -4,6 +4,7 @@
|
|||
"fallback": [
|
||||
"kimi-coding",
|
||||
"qwen-code",
|
||||
"google-cloud",
|
||||
"gemini",
|
||||
"anthropic",
|
||||
"openai",
|
||||
|
|
@ -177,9 +178,24 @@
|
|||
"fallback": true,
|
||||
"notes": ""
|
||||
},
|
||||
{
|
||||
"id": "google-cloud",
|
||||
"label": "Google Cloud (Vertex AI)",
|
||||
"kind": "api",
|
||||
"protocol": "gemini",
|
||||
"auth": "api_key",
|
||||
"keyEnv": "GOOGLE_CLOUD_API_KEY",
|
||||
"baseUrl": "https://us-central1-aiplatform.googleapis.com/v1",
|
||||
"defaultModel": "gemini-2.5-pro",
|
||||
"embeddingModel": "text-embedding-004",
|
||||
"maxContext": 1000000,
|
||||
"streaming": true,
|
||||
"fallback": true,
|
||||
"notes": "Vertex AI enterprise endpoint. OAuth/service-account fallback supported by GOOGLE_APPLICATION_CREDENTIALS."
|
||||
},
|
||||
{
|
||||
"id": "gemini",
|
||||
"label": "Google Gemini",
|
||||
"label": "Google Gemini (AI Studio)",
|
||||
"kind": "api",
|
||||
"protocol": "gemini",
|
||||
"auth": "api_key",
|
||||
|
|
|
|||
|
|
@ -198,8 +198,24 @@ providers:
|
|||
streaming: true
|
||||
fallback: true
|
||||
|
||||
# ── Google Cloud ────────────────────────────────────────────────────────────
|
||||
- id: google-cloud
|
||||
label: Google Cloud (Vertex AI)
|
||||
kind: api
|
||||
auth: api_key
|
||||
key_env: GOOGLE_CLOUD_API_KEY
|
||||
base_url: https://us-central1-aiplatform.googleapis.com/v1
|
||||
protocol: gemini
|
||||
chat_model: gemini-2.5-pro
|
||||
embedding_model: text-embedding-004
|
||||
max_context: 1000000
|
||||
surfaces: [dirac, mcp, portal]
|
||||
streaming: true
|
||||
fallback: true
|
||||
notes: Vertex AI enterprise endpoint. OAuth/service-account fallback supported by GOOGLE_APPLICATION_CREDENTIALS.
|
||||
|
||||
- id: gemini
|
||||
label: Google Gemini
|
||||
label: Google Gemini (AI Studio)
|
||||
kind: api
|
||||
auth: api_key
|
||||
key_env: GEMINI_API_KEY
|
||||
|
|
@ -340,6 +356,7 @@ providers:
|
|||
mcp_chat_fallback:
|
||||
- kimi-coding
|
||||
- qwen-code
|
||||
- google-cloud
|
||||
- gemini
|
||||
- anthropic
|
||||
- openai
|
||||
|
|
@ -349,6 +366,7 @@ mcp_chat_fallback:
|
|||
|
||||
# Embedding preference order for the Python MCP router.
|
||||
mcp_embed_fallback:
|
||||
- google-cloud
|
||||
- gemini
|
||||
- openai
|
||||
- qwen
|
||||
|
|
|
|||
220
.aurelio/skills/gcp/SKILL.md
Normal file
220
.aurelio/skills/gcp/SKILL.md
Normal file
|
|
@ -0,0 +1,220 @@
|
|||
---
|
||||
name: gcp
|
||||
description: Google Cloud platform operations for PF — project setup, Vertex AI models, GKE, Cloud Run, Cloud Functions, Cloud Storage, Pub/Sub, BigQuery, Cloud SQL, Secret Manager, IAM, service accounts, and domain-wide delegation. Use when the user needs GCP-specific architecture, provisioning, or operations; prefer self-hosted or emulated dev paths before touching real GCP projects.
|
||||
---
|
||||
|
||||
# gcp — Google Cloud Platform (enterprise)
|
||||
|
||||
Distinct from the `gemini` provider (AI Studio consumer API) and `cli-gcloud`
|
||||
(gcloud command snippets). This skill covers **Google Cloud Vertex AI / enterprise
|
||||
workloads**: project setup, billing, IAM, model serving, compute, data, messaging,
|
||||
and secrets.
|
||||
|
||||
## When to use
|
||||
|
||||
- New client/project onboarding to GCP
|
||||
- Vertex AI model selection and routing
|
||||
- Deploying or operating GKE, Cloud Run, Cloud Functions
|
||||
- Cloud Storage, Pub/Sub, BigQuery, Cloud SQL, Secret Manager design
|
||||
- Service accounts, workload identity, domain-wide delegation
|
||||
- Deciding when to use GCP vs self-hosted PF infrastructure
|
||||
|
||||
## When NOT to use
|
||||
|
||||
- Consumer Gemini (AI Studio) → use the `gemini` provider in `.aurelio/providers/registry.yaml`
|
||||
- Simple gcloud CLI one-liners → use `cli-gcloud` skill
|
||||
- Firebase/GCP mobile-only features → not covered here
|
||||
|
||||
## Project setup checklist
|
||||
|
||||
1. **Create/select project**
|
||||
```bash
|
||||
gcloud projects create pf-<client>-<env> --name="PF <Client> <Env>"
|
||||
gcloud config set project pf-<client>-<env>
|
||||
```
|
||||
|
||||
2. **Billing**
|
||||
- Link billing account in Cloud Console or via `gcloud billing projects link`
|
||||
- Set budget alerts (recommended: 50 %, 90 %, 100 %)
|
||||
- For clients, tag resources (`cost-center`, `client`, `realm`)
|
||||
|
||||
3. **Enable APIs** (one-time per project)
|
||||
```bash
|
||||
gcloud services enable aiplatform.googleapis.com
|
||||
gcloud services enable container.googleapis.com
|
||||
gcloud services enable run.googleapis.com
|
||||
gcloud services enable cloudfunctions.googleapis.com
|
||||
gcloud services enable storage.googleapis.com
|
||||
gcloud services enable pubsub.googleapis.com
|
||||
gcloud services enable bigquery.googleapis.com
|
||||
gcloud services enable sqladmin.googleapis.com
|
||||
gcloud services enable secretmanager.googleapis.com
|
||||
gcloud services enable cloudbuild.googleapis.com
|
||||
gcloud services enable iamcredentials.googleapis.com
|
||||
```
|
||||
|
||||
4. **IAM baseline**
|
||||
- Create admin group `gcp-pf-admins@<domain>` with `roles/owner` or `roles/editor`
|
||||
- Assign `roles/viewer` to auditors
|
||||
- Enable essential audit logs (Admin Activity + Data Access for Secret Manager / Cloud SQL)
|
||||
|
||||
## Service accounts & auth
|
||||
|
||||
- **Human users**: use `gcloud auth login` + IAM bindings; no service-account keys in laptops.
|
||||
- **Workloads**: use Workload Identity Federation where possible; otherwise create
|
||||
dedicated service accounts per workload.
|
||||
- **Keys**: store JSON keys in Vaultwarden / Secret Manager; rotate every 90 days.
|
||||
- **Domain-wide delegation**: required for Workspace APIs (Directory, Gmail, Calendar, Drive).
|
||||
- GCP Console → IAM & Admin → Service Accounts → `pf-<workload>@<project>.iam.gserviceaccount.com`
|
||||
- Add domain-wide delegation; note the Client ID
|
||||
- Google Admin Console → Security → API controls → Domain-wide delegation
|
||||
→ Authorize the Client ID with OAuth scopes needed
|
||||
- Store Client ID and delegated SA email in Secret Manager
|
||||
|
||||
## Vertex AI
|
||||
|
||||
Use Vertex AI for enterprise Gemini, Imagen, Veo, and embeddings. Region choice
|
||||
affects latency and model availability; `europe-west1` is PF's default for EU
|
||||
data residency.
|
||||
|
||||
### Model IDs
|
||||
|
||||
| Model | ID | Context | Type | Notes |
|
||||
|-------|----|---------|------|-------|
|
||||
| Gemini 2.5 Pro | `gemini-2.5-pro-preview-06-05` | 1M tokens | chat | Reasoning, long context, code |
|
||||
| Gemini 2.5 Flash | `gemini-2.5-flash-preview-06-05` | 1M tokens | chat | Fast, cheaper |
|
||||
| Gemini 2.0 Flash | `gemini-2.0-flash-001` | 1M tokens | chat | Stable GA |
|
||||
| Imagen 3 | `imagen-3-0-generate-001` | N/A | image | Text-to-image |
|
||||
| Veo 2 | `veo-2-0-generate-001` | N/A | video | Text/video-to-video |
|
||||
| Text Embedding 004 | `text-embedding-004` | 2048 dims | embedding | Sentence + document |
|
||||
| Multimodal Embedding | `multimodalembedding@001` | 128/256/512/1408 dims | embedding | Image + text |
|
||||
|
||||
### Endpoint pattern
|
||||
|
||||
```
|
||||
https://<region>-aiplatform.googleapis.com/v1/projects/<project>/locations/<region>/publishers/google/models/<model-id>
|
||||
```
|
||||
|
||||
### Example: generate content
|
||||
|
||||
```bash
|
||||
REGION=europe-west1
|
||||
PROJECT=$(gcloud config get-value project)
|
||||
MODEL=gemini-2.5-flash-preview-06-05
|
||||
|
||||
curl -X POST \
|
||||
"https://${REGION}-aiplatform.googleapis.com/v1/projects/${PROJECT}/locations/${REGION}/publishers/google/models/${MODEL}:generateContent" \
|
||||
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{"contents":[{"role":"user","parts":[{"text":"Summarize this in one line."}]}]}'
|
||||
```
|
||||
|
||||
### Python (google-cloud-aiplatform)
|
||||
|
||||
```python
|
||||
import vertexai
|
||||
from vertexai.generative_models import GenerativeModel
|
||||
|
||||
vertexai.init(project="pf-client-env", location="europe-west1")
|
||||
model = GenerativeModel("gemini-2.5-flash-preview-06-05")
|
||||
response = model.generate_content("Hello, GCP world.")
|
||||
print(response.text)
|
||||
```
|
||||
|
||||
## Compute
|
||||
|
||||
### GKE
|
||||
|
||||
- Use Autopilot for most PF workloads unless node-level tuning is required.
|
||||
- Default region: `europe-west1`; multi-region only when HA is justified.
|
||||
- Enable Workload Identity; disable legacy metadata endpoints.
|
||||
|
||||
```bash
|
||||
gcloud container clusters create-auto pf-cluster \
|
||||
--region=europe-west1 \
|
||||
--release-channel=regular \
|
||||
--enable-workload-identity
|
||||
```
|
||||
|
||||
### Cloud Run
|
||||
|
||||
- Best for stateless HTTP services and MCP servers
|
||||
- Use `--no-allow-unauthenticated` for internal agents; use IAM/service accounts for access
|
||||
- Connect to Cloud SQL via proxy / native connections
|
||||
|
||||
```bash
|
||||
gcloud run deploy pf-agent-service \
|
||||
--source . \
|
||||
--region=europe-west1 \
|
||||
--no-allow-unauthenticated \
|
||||
--service-account=pf-run-sa@$PROJECT.iam.gserviceaccount.com
|
||||
```
|
||||
|
||||
### Cloud Functions
|
||||
|
||||
- Use 2nd gen for longer timeouts, bigger instances, and Eventarc triggers
|
||||
- Trigger from Pub/Sub, Cloud Storage, Firestore, HTTP
|
||||
|
||||
## Data & messaging
|
||||
|
||||
### Cloud Storage
|
||||
|
||||
- Buckets per realm/env; uniform bucket-level access enabled
|
||||
- Lifecycle rules for logs and artifacts
|
||||
- `gsutil` / `gcloud storage` are interchangeable; prefer `gcloud storage`
|
||||
|
||||
```bash
|
||||
gcloud storage buckets create gs://pf-client-data-euw1 --location=europe-west1
|
||||
gcloud storage buckets update gs://pf-client-data-euw1 --uniform-bucket-level-access
|
||||
```
|
||||
|
||||
### Pub/Sub
|
||||
|
||||
- Default for async agent-to-agent messaging
|
||||
- Use push subscriptions to Cloud Run for reactive agents
|
||||
- Enable exactly-once delivery when order/correctness matters
|
||||
|
||||
```bash
|
||||
gcloud pubsub topics create agent-events
|
||||
gcloud pubsub subscriptions create agent-events-run \
|
||||
--topic agent-events \
|
||||
--push-endpoint=https://pf-agent-service-xxx-uc.a.run.app/events
|
||||
```
|
||||
|
||||
### BigQuery
|
||||
|
||||
- Use for analytics, telemetry, and structured agent memory at scale
|
||||
- Datasets per client/realm; partition large tables by ingestion time
|
||||
- Prefer service-account access via IAM `roles/bigquery.dataViewer`
|
||||
|
||||
### Cloud SQL
|
||||
|
||||
- PostgreSQL 16 default; use Cloud SQL Auth Proxy for local/dev access
|
||||
- Private IP + VPC connector for GKE/Cloud Run
|
||||
- Backups and point-in-time recovery enabled for production
|
||||
|
||||
## Secret Manager
|
||||
|
||||
- Central store for API keys, service-account JSONs, database passwords
|
||||
- Name convention: `<client>/<realm>/<secret>`
|
||||
- Rotate via Secret Manager versions; latest alias for runtime
|
||||
|
||||
```bash
|
||||
gcloud secrets create pf-client-db-password --data-file=-
|
||||
# then paste value, Ctrl+D
|
||||
```
|
||||
|
||||
## Surfaces
|
||||
|
||||
- **dirac**: Vertex AI models appear as provider `google-cloud` in provider registry
|
||||
- **mcp**: MCP servers running on Cloud Run / GKE; `model_router.py` can route to Vertex AI
|
||||
- **portal**: PF portal can expose Vertex AI models and GCP-hosted agents under the Google Cloud offering
|
||||
|
||||
## Pitfalls
|
||||
|
||||
- `gemini` (AI Studio) and `google-cloud` (Vertex AI) are separate providers; never ship a Vertex AI key to an AI Studio endpoint.
|
||||
- GCP IAM is eventually consistent; wait or retry after policy changes.
|
||||
- Region matters for model availability and pricing — verify in Vertex AI Model Garden before hard-coding a region.
|
||||
- Domain-wide delegation is powerful; limit scopes and monitor audit logs.
|
||||
- Cloud Run cold starts can add latency; use min-instances for latency-sensitive agents.
|
||||
- Storage egress and BigQuery query costs can surprise; set budgets early.
|
||||
39
.env.example
39
.env.example
|
|
@ -60,3 +60,42 @@ TELEGRAM_TOKEN=
|
|||
TELEGRAM_ADMINS=
|
||||
# Optional sandbox root for `cd` / file transfers (absolute path).
|
||||
TELEGRAM_WORKING_DIR=/home/hermes/telminal-workspace
|
||||
|
||||
# ─── MIRROR PIPELINE (sync-mirrors.py) ──────────────────────────────────────
|
||||
# Upstream sources replicate into self-hosted mirrors:
|
||||
# GitHub → Forgejo (code)
|
||||
# Jira Cloud → Plane (issues)
|
||||
# Confluence → Outline (docs)
|
||||
#
|
||||
# Registry: .aurelio/mirrors/sync-mirrors.yaml (edit THIS)
|
||||
# Run: python3 scripts/sync-mirrors.py --sync
|
||||
#
|
||||
# ALL tokens are API tokens/PATs — fetch from Vaultwarden, never hardcode.
|
||||
|
||||
# Forgejo (mirror target) — admin token with repo + org scopes.
|
||||
# Generate: Settings → Applications → Access Tokens (manage_repo, write:org)
|
||||
FORGEJO_MIRROR_TOKEN=
|
||||
|
||||
# GitHub (upstream source) — classic PAT with repo + read:org scopes.
|
||||
# Generate: https://github.com/settings/tokens (classic)
|
||||
GITHUB_MIRROR_TOKEN=
|
||||
|
||||
# Plane (mirror target) — personal API token.
|
||||
# Generate: Profile → Settings → API Tokens
|
||||
PLANE_API_TOKEN=
|
||||
|
||||
# Outline (mirror target) — API token with read+write.
|
||||
# Generate: Settings → API Tokens
|
||||
OUTLINE_API_TOKEN=
|
||||
|
||||
# Jira Cloud (upstream source) — API token + account email.
|
||||
# Generate: https://id.atlassian.com/manage-profile/security/api-tokens
|
||||
JIRA_API_TOKEN=
|
||||
JIRA_USER_EMAIL=
|
||||
# Jira Cloud base URL (e.g. https://savearth.atlassian.net) — set in sync-mirrors.yaml
|
||||
# Leave empty here; the YAML endpoint field is null until the instance exists.
|
||||
|
||||
# Confluence Cloud (upstream source) — same Atlassian account as Jira.
|
||||
# Generate: same path as Jira (shares the Atlassian account)
|
||||
CONFLUENCE_API_TOKEN=
|
||||
CONFLUENCE_USER_EMAIL=
|
||||
|
|
|
|||
3
.gitignore
vendored
3
.gitignore
vendored
|
|
@ -64,3 +64,6 @@ hetzner.pub
|
|||
id_rsa
|
||||
id_ed25519
|
||||
|
||||
.runner
|
||||
.mimocode/.cron-lock
|
||||
infrastructure/fabric/drift/target/
|
||||
|
|
|
|||
240
.hermes/plans/2026-07-14_160000-music-mvp-aurelio-telegram.md
Normal file
240
.hermes/plans/2026-07-14_160000-music-mvp-aurelio-telegram.md
Normal file
|
|
@ -0,0 +1,240 @@
|
|||
# Music MVP — Full Plan (Debug Mode + Daily Overview + Service Integration)
|
||||
|
||||
> **For Hermes:** Use subagent-driven-development skill to implement this plan task-by-task.
|
||||
|
||||
**Goal:** A unified music intelligence layer accessible from Telegram. Debug mode for service visibility, daily music digest, and deep integration with the full media stack.
|
||||
|
||||
---
|
||||
|
||||
## Services Map (Gigabyte — 192.168.0.104)
|
||||
|
||||
| Service | Port | Docker Container | API | Auth | Status |
|
||||
|---------|------|-----------------|-----|------|--------|
|
||||
| **Navidrome** | 4533 | pf-navidrome | Subsonic REST | user `fabio`, md5 token+salt | 🟢 |
|
||||
| **Lidarr** | 8686 | pf-lidarr | `/api/v1` | `X-Api-Key: 24077d03b9a742a786dd799567659cb8` | 🟢 688 artists |
|
||||
| **Prowlarr** | 9696 | pf-prowlarr | `/api/v1` | `X-Api-Key: c45adb4eff734ea491d04e1cbe93bca9` | 🟢 |
|
||||
| **Slskd** | 5030 | pf-slskd | `/api/v0` | TBD | 🟢 v0.25.1 |
|
||||
| **Jellyfin** | 8096 | pf-jellyfin | REST | API key `bc8c6c069daf4e4c8fec660a92114ac4` | 🟢 |
|
||||
| **Maloja** | 42010 | CT 216 | `/apis/mlj_1/` | public (charts) | 🟢 |
|
||||
| **Multi-scrobbler** | 9078 | CT 216 | — | — | 🟢 |
|
||||
| **qBittorrent** | — | pf-qbittorrent | REST | TBD | 🟢 |
|
||||
|
||||
### External Services
|
||||
|
||||
| Service | API | Auth | Purpose |
|
||||
|---------|-----|------|---------|
|
||||
| **Last.fm** | `ws.audioscrobbler.com` | API key `b7bc626626486b5b60555db6ef836976`, session `L2nx0sc...` | Scrobbling + charts |
|
||||
| **MusicBrainz** | `musicbrainz.org/ws/2` | User-Agent header | Artist/album metadata, MBIDs |
|
||||
| **ListenBrainz** | `api.listenbrainz.org/1` | Token TBD | Open-source scrobbling + recommendations |
|
||||
|
||||
### Data Flow
|
||||
|
||||
```
|
||||
Playing music:
|
||||
Navidrome/Jellyfin → Multi-scrobbler → Maloja + Last.fm + [ListenBrainz]
|
||||
|
||||
Discovering music:
|
||||
Lidarr → Prowlarr (indexers) → qBittorrent → library
|
||||
MusicBrainz → artist metadata, MBIDs, release groups
|
||||
Last.fm → user charts, similar artists, tags
|
||||
ListenBrainz → recommendations, fresh releases
|
||||
|
||||
Converting music:
|
||||
Navidrome (download) → MuScriptor (transcribe) → MIDI files
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Feature 1: Debug Mode
|
||||
|
||||
### What the user sees in Telegram:
|
||||
|
||||
```
|
||||
/debug music — full status dashboard
|
||||
/debug music navidrome — Navidrome logs + config
|
||||
/debug music lidarr — Lidarr queue + recent activity
|
||||
/debug music scrobbler — Multi-scrobbler status + recent scrobbles
|
||||
/debug music services — all service health at a glance
|
||||
/debug music config navidrome — Navidrome config dump
|
||||
```
|
||||
|
||||
### Tools the MCP server exposes:
|
||||
|
||||
```python
|
||||
# Debug & Monitoring
|
||||
music_debug_services() # health check all services
|
||||
music_debug_logs(service, lines=50) # tail Docker logs
|
||||
music_debug_config(service) # dump service config
|
||||
music_debug_scrobbler_status() # multi-scrobbler sources + clients
|
||||
music_debug_lidarr_activity() # recent Lidarr activity (imports, grabs)
|
||||
music_debug_disk_usage() # music library disk usage
|
||||
```
|
||||
|
||||
### Implementation:
|
||||
|
||||
Each debug tool maps to a Docker API call or service REST endpoint:
|
||||
- `docker logs --tail N <container>` for logs
|
||||
- `docker inspect <container>` for config/env
|
||||
- Service REST APIs for status endpoints
|
||||
|
||||
---
|
||||
|
||||
## Feature 2: Daily Music Overview
|
||||
|
||||
### What the user sees every morning:
|
||||
|
||||
```
|
||||
🎵 Daily Music Brief — Monday, July 14
|
||||
|
||||
📊 Listening Stats (Last.fm/Maloja)
|
||||
Top this week: LCD Soundsystem (12 plays), Caribou (8), Daft Punk (5)
|
||||
Total scrobbles today: 23
|
||||
New discovery: "Komori Uta" by YIN YIN
|
||||
|
||||
🆕 New Releases (from your 688 Lidarr artists)
|
||||
• Caribou — "Honey" (single, released today)
|
||||
• Tame Impala — "Wings of Time" (single, 2 days ago)
|
||||
• LCD Soundsystem — "new body rhumba" (single, 3 days ago)
|
||||
|
||||
📥 Library Status
|
||||
Lidarr: 2 albums downloading, 0 missing
|
||||
Disk: 1.2 TB / 2 TB (60%)
|
||||
|
||||
🔗 Scrobbler Health
|
||||
Navidrome → Maloja: ✅ Last: 2 min ago
|
||||
Jellyfin → Maloja: ✅ Last: 1h ago
|
||||
Navidrome → Last.fm: ✅ Last: 2 min ago
|
||||
```
|
||||
|
||||
### Tools:
|
||||
|
||||
```python
|
||||
# Daily Overview
|
||||
music_daily_brief() # full daily report
|
||||
music_new_releases(days=7) # new releases from followed artists
|
||||
music_listening_stats(period="week") # Last.fm + Maloja stats
|
||||
music_library_health() # disk, missing, downloading
|
||||
```
|
||||
|
||||
### New Releases Detection:
|
||||
|
||||
Two approaches combined:
|
||||
1. **Lidarr calendar** — `GET /api/v1/calendar?start=...&end=...` — releases Lidarr knows about
|
||||
2. **MusicBrainz** — `GET /ws/2/release-group?artist=<mbid>&type=album|single&first-release-date:[NOW-7DAYS TO NOW]` — any new release from followed artists
|
||||
3. **Last.fm** — `artist.getNewReleases` — if available
|
||||
|
||||
---
|
||||
|
||||
## Feature 3: ListenBrainz Integration
|
||||
|
||||
### Why:
|
||||
- Open-source alternative to Last.fm
|
||||
- Better recommendation engine
|
||||
- Cross-compatible with MusicBrainz MBIDs
|
||||
- Can run alongside Last.fm (multi-scrobbler supports multiple clients)
|
||||
|
||||
### How:
|
||||
1. Register at `listenbrainz.org`
|
||||
2. Get API token
|
||||
3. Add ListenBrainz as a client in multi-scrobbler config
|
||||
4. Use ListenBrainz API for recommendations + fresh releases
|
||||
|
||||
### Multi-scrobbler config addition:
|
||||
|
||||
```json
|
||||
{
|
||||
"type": "listenbrainz",
|
||||
"name": "ListenBrainz",
|
||||
"data": {
|
||||
"url": "https://api.listenbrainz.org",
|
||||
"token": "<LISTENBRAINZ_TOKEN>"
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Feature 4: MusicBrainz Metadata
|
||||
|
||||
### What MusicBrainz provides:
|
||||
- Canonical artist/album/track IDs (MBIDs)
|
||||
- Release groups, release dates, track listings
|
||||
- Artist relationships (collaborations, aliases)
|
||||
- Genre/tag taxonomy
|
||||
- Cover art via Cover Art Archive
|
||||
|
||||
### Tools:
|
||||
|
||||
```python
|
||||
# MusicBrainz
|
||||
music_brainz_search(query) # search for artist/album
|
||||
music_brainz_artist_info(mbid) # detailed artist info
|
||||
music_brainz_releases(mbid, days=30) # recent releases for artist
|
||||
music_brainz_similar(mbid) # similar artists
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Implementation Tasks
|
||||
|
||||
### Task 1: music-mcp server skeleton (with debug tools)
|
||||
Same pattern as muscriptor-mcp. FastMCP SSE on port 8089.
|
||||
|
||||
### Task 2: Navidrome client (search, browse, download, stream)
|
||||
Subsonic API: `search3`, `getAlbumList2`, `getRandomSongs`, `getNowPlaying`, `stream`, `download`.
|
||||
|
||||
### Task 3: Lidarr client (library management + calendar)
|
||||
Lidarr v3 API: artist CRUD, queue, wanted, calendar, lookup.
|
||||
|
||||
### Task 4: Prowlarr client (indexer search)
|
||||
`GET /api/v1/search` — search across all configured indexers.
|
||||
|
||||
### Task 5: Last.fm client (stats, charts, new releases)
|
||||
`user.getTopArtists`, `user.getTopTracks`, `user.getRecentTracks`, `artist.getInfo`, `artist.getSimilar`.
|
||||
|
||||
### Task 6: MusicBrainz client (metadata, release groups)
|
||||
`/ws/2/artist`, `/ws/2/release-group`, `/ws/2/release` with proper User-Agent.
|
||||
|
||||
### Task 7: Debug tools (logs, config, health)
|
||||
Docker API calls for container logs + inspect. Service REST endpoints for status.
|
||||
|
||||
### Task 8: Daily overview composer
|
||||
Combines data from all clients into a formatted daily brief.
|
||||
|
||||
### Task 9: MuScriptor bridge
|
||||
Calls muscriptor-mcp's `transcribe_to_midi` tool.
|
||||
|
||||
### Task 10: Register + deploy
|
||||
Register in `.aurelio/mcp_config.json`. Deploy on Gigabyte. Set up daily cron.
|
||||
|
||||
### Task 11: ListenBrainz setup
|
||||
Register account, add to multi-scrobbler, verify scrobbling works.
|
||||
|
||||
---
|
||||
|
||||
## Files Summary
|
||||
|
||||
| Action | Path |
|
||||
|--------|------|
|
||||
| Create | `scripts/music-mcp/pyproject.toml` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/__init__.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/server.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/navidrome_client.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/lidarr_client.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/prowlarr_client.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/lastfm_client.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/musicbrainz_client.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/muscriptor_bridge.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/debug_tools.py` |
|
||||
| Create | `scripts/music-mcp/src/music_mcp/daily_overview.py` |
|
||||
| Create | `scripts/music-mcp/systemd/music-mcp.service` |
|
||||
| Modify | `.aurelio/mcp_config.json` |
|
||||
|
||||
## Cron Schedule
|
||||
|
||||
| Time | Job | Output |
|
||||
|------|-----|--------|
|
||||
| 8:00 AM | `music_daily_brief()` | Telegram: daily music overview |
|
||||
| On-demand | `/debug music` | Telegram: service status |
|
||||
| On-demand | `/search Caribou` | Telegram: unified search results |
|
||||
| On-demand | `/midi Sun Caribou` | Telegram: MIDI conversion |
|
||||
578
.hermes/plans/2026-07-14_180000-muscriptor-mcp-weekly-midi.md
Normal file
578
.hermes/plans/2026-07-14_180000-muscriptor-mcp-weekly-midi.md
Normal file
|
|
@ -0,0 +1,578 @@
|
|||
# MuScriptor MCP Server + Weekly MIDI Pipeline
|
||||
|
||||
> **For Hermes:** Use subagent-driven-development skill to implement this plan task-by-task.
|
||||
|
||||
**Goal:** Build a MuScriptor-powered MCP server that agents (Hermes, Aurelio) can call to transcribe audio → MIDI, and a weekly cron that auto-converts the user's top 3 tracks from their Navidrome/Maloja music stack.
|
||||
|
||||
**Architecture:** Python MCP server exposing `transcribe_to_midi`, `get_top_tracks`, `search_and_transcribe` tools. Backed by MuScriptor (Kyutai/Mirelo, 1B param music transcription transformer). Connects to existing Maloja (stats) + Navidrome (audio source) via their APIs. Deployed as a systemd service on Gigabyte (RTX 3070 GPUs) or Tomahawk MAX. Weekly Hermes cron job queries top tracks → downloads → transcribes → delivers MIDI.
|
||||
|
||||
**Tech Stack:** Python 3.12+, MuScriptor (`pip install muscriptor`), MCP Python SDK, Maloja API, Navidrome Subsonic API, Hermes cron.
|
||||
|
||||
---
|
||||
|
||||
## Context
|
||||
|
||||
### Existing Music Infrastructure
|
||||
|
||||
| Service | URL | Port | Host | Purpose |
|
||||
|---------|-----|------|------|---------|
|
||||
| Navidrome | `music.portugalfuturista.org` | 4533 | Tomahawk MAX (Gigabyte) Docker | Music library + Subsonic API |
|
||||
| Maloja | `maloja.portugalfuturista.org` | 42010 | CT 216 | Scrobble statistics |
|
||||
| Multi-scrobbler | `scrobbler.portugalfuturista.org` | 9078 | CT 216 | Aggregates from Jellyfin, Navidrome, YTMusic → Maloja + Last.fm |
|
||||
| Jellyfin | — | 8096 | Gigabyte Docker | Media server (also scrobbles) |
|
||||
|
||||
### MuScriptor Model (Kyutai + Mirelo)
|
||||
|
||||
- **Repo:** `github.com/muscriptor/muscriptor` (MIT license)
|
||||
- **Models:** `muscriptor-small` (103M), `muscriptor-medium` (307M, default), `muscriptor-large` (1.4B)
|
||||
- **License:** CC BY-NC 4.0 (requires free HF account + token)
|
||||
- **Install:** `pip install muscriptor` or `uvx muscriptor`
|
||||
- **Key API:** `model.transcribe_to_midi("audio.wav")` → MIDI bytes
|
||||
- **GPU:** medium runs fine on RTX 3070 (8GB); large needs ~12GB VRAM
|
||||
- **Supported instruments:** piano, drums, guitar, bass, strings, winds, etc.
|
||||
|
||||
### How It Connects to Aurelio / Hermes
|
||||
|
||||
- MCP server registered in `.aurelio/mcp_config.json` → available to all IDE agents
|
||||
- Hermes cron job calls the MCP tools weekly
|
||||
- User can also invoke directly: *"Hermes, convert Sun by Caribou to MIDI"*
|
||||
|
||||
---
|
||||
|
||||
## Implementation Plan
|
||||
|
||||
### Phase 1: MCP Server Core (muscriptIdor-mcp)
|
||||
|
||||
#### Task 1: Create MCP server skeleton
|
||||
|
||||
**Objective:** Scaffold the Python MCP server project structure.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/pyproject.toml`
|
||||
- Create: `scripts/muscriptor-mcp/src/muscriptor_mcp/__init__.py`
|
||||
- Create: `scripts/muscriptor-mcp/src/muscriptor_mcp/server.py`
|
||||
|
||||
**Step 1:** Create project with `pyproject.toml`:
|
||||
|
||||
```toml
|
||||
[project]
|
||||
name = "muscriptor-mcp"
|
||||
version = "0.1.0"
|
||||
description = "MCP server for MuScriptor music transcription (audio → MIDI)"
|
||||
requires-python = ">=3.10"
|
||||
dependencies = [
|
||||
"mcp[cli]>=1.0.0",
|
||||
"muscriptor>=0.1.0",
|
||||
"httpx>=0.27",
|
||||
"pydantic>=2.0",
|
||||
]
|
||||
|
||||
[project.scripts]
|
||||
muscriptor-mcp = "muscriptor_mcp.server:main"
|
||||
```
|
||||
|
||||
**Step 2:** Create the MCP server skeleton with `server.py`:
|
||||
|
||||
```python
|
||||
"""MuScriptor MCP Server — audio → MIDI transcription for Aurelio agents."""
|
||||
import asyncio
|
||||
import logging
|
||||
from pathlib import Path
|
||||
from mcp.server import Server
|
||||
from mcp.server.stdio import run_server
|
||||
from mcp.types import Tool, TextContent
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
app = Server("muscriptor-mcp")
|
||||
|
||||
@app.list_tools()
|
||||
async def list_tools() -> list[Tool]:
|
||||
return [
|
||||
Tool(
|
||||
name="transcribe_to_midi",
|
||||
description="Transcribe an audio file (WAV/MP3/FLAC/OGG) to MIDI using MuScriptor. "
|
||||
"Returns the path to the generated .mid file.",
|
||||
inputSchema={
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"audio_path": {"type": "string", "description": "Path to audio file"},
|
||||
"output_dir": {"type": "string", "description": "Output directory for MIDI (default: same as input)"},
|
||||
"model_size": {"type": "string", "enum": ["small", "medium", "large"], "default": "medium"},
|
||||
"instruments": {"type": "array", "items": {"type": "string"}, "description": "Restrict to specific instruments"},
|
||||
},
|
||||
"required": ["audio_path"],
|
||||
},
|
||||
),
|
||||
Tool(
|
||||
name="get_top_tracks",
|
||||
description="Get the user's most-played tracks from Maloja scrobble stats.",
|
||||
inputSchema={
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"count": {"type": "integer", "default": 3, "description": "Number of top tracks"},
|
||||
"period": {"type": "string", "default": "week", "description": "Time period: day, week, month, year, overall"},
|
||||
},
|
||||
},
|
||||
),
|
||||
Tool(
|
||||
name="search_and_download",
|
||||
description="Search Navidrome for a track and download the audio file locally.",
|
||||
inputSchema={
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"query": {"type": "string", "description": "Search query (e.g. 'Sun Caribou')"},
|
||||
"download_dir": {"type": "string", "description": "Directory to save audio (default: /tmp/muscriptor)"},
|
||||
},
|
||||
"required": ["query"],
|
||||
},
|
||||
),
|
||||
Tool(
|
||||
name="top_tracks_to_midi",
|
||||
description="Full pipeline: fetch top N tracks from Maloja, download from Navidrome, "
|
||||
"transcribe each to MIDI. Returns paths to all generated .mid files.",
|
||||
inputSchema={
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"count": {"type": "integer", "default": 3},
|
||||
"period": {"type": "string", "default": "week"},
|
||||
"model_size": {"type": "string", "enum": ["small", "medium", "large"], "default": "medium"},
|
||||
},
|
||||
},
|
||||
),
|
||||
]
|
||||
|
||||
async def main():
|
||||
async with run_server(app) as server:
|
||||
await server.serve_forever()
|
||||
|
||||
if __name__ == "__main__":
|
||||
asyncio.run(main())
|
||||
```
|
||||
|
||||
**Verification:** `cd scripts/muscriptor-mcp && pip install -e . && python -c "from muscriptor_mcp.server import app; print('OK')"`
|
||||
|
||||
---
|
||||
|
||||
#### Task 2: Implement Maloja API client (get_top_tracks)
|
||||
|
||||
**Objective:** Query Maloja for top tracks with play counts.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/src/muscriptor_mcp/maloja_client.py`
|
||||
|
||||
**Step 1:** Maloja exposes a REST API. Key endpoints:
|
||||
|
||||
```
|
||||
GET /api/v1/top/tracks?period=week&max=3
|
||||
Authorization: Token <api_key>
|
||||
```
|
||||
|
||||
Response shape (from Maloja docs):
|
||||
```json
|
||||
{
|
||||
"list": [
|
||||
{
|
||||
"track": {"title": "Sun", "artists": [{"name": "Caribou"}]},
|
||||
"amount": 12
|
||||
}
|
||||
]
|
||||
}
|
||||
```
|
||||
|
||||
**Step 2:** Implement the client:
|
||||
|
||||
```python
|
||||
"""Maloja scrobble statistics client."""
|
||||
import httpx
|
||||
from dataclasses import dataclass
|
||||
|
||||
MALOJA_URL = "http://pf-maloja:42010" # LAN — also accessible via maloja.portugalfuturista.org
|
||||
MALOJA_API_KEY = "" # TODO: load from env MALOJA_API_KEY
|
||||
|
||||
@dataclass
|
||||
class TrackStat:
|
||||
title: str
|
||||
artist: str
|
||||
play_count: int
|
||||
|
||||
async def get_top_tracks(count: int = 3, period: str = "week") -> list[TrackStat]:
|
||||
"""Fetch top tracks from Maloja API."""
|
||||
url = f"{MALOJA_URL}/api/v1/top/tracks"
|
||||
params = {"period": period, "max": count}
|
||||
headers = {"Authorization": f"Token {MALOJA_API_KEY}"}
|
||||
async with httpx.AsyncClient() as client:
|
||||
resp = await client.get(url, params=params, headers=headers, timeout=15)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
return [
|
||||
TrackStat(
|
||||
title=item["track"]["title"],
|
||||
artist=item["track"]["artists"][0]["name"],
|
||||
play_count=item["amount"],
|
||||
)
|
||||
for item in data["list"]
|
||||
]
|
||||
```
|
||||
|
||||
**Verification:** Test with `curl` against live Maloja first to confirm API shape, then unit test with mocked response.
|
||||
|
||||
---
|
||||
|
||||
#### Task 3: Implement Navidrome Subsonic API client (search_and_download)
|
||||
|
||||
**Objective:** Search Navidrome and download audio files.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/src/muscriptor_mcp/navidrome_client.py`
|
||||
|
||||
**Step 1:** Navidrome exposes the Subsonic API. Key endpoints:
|
||||
|
||||
```
|
||||
GET /rest/search3.view?query=Sun+Caribou&u=fabio&t=<token>&s=<salt>&v=1.16.1&c=muscriptor
|
||||
GET /rest/download.view?id=<songId>&u=fabio&t=<token>&s=<salt>&v=1.16.1&c=muscriptor
|
||||
```
|
||||
|
||||
Subsonic auth: token = md5(password + salt).
|
||||
|
||||
**Step 2:** Implement:
|
||||
|
||||
```python
|
||||
"""Navidrome (Subsonic API) client for audio download."""
|
||||
import hashlib
|
||||
import os
|
||||
import secrets
|
||||
from pathlib import Path
|
||||
import httpx
|
||||
|
||||
NAVIDROME_URL = "http://pf-navidrome:4533" # LAN
|
||||
NAVIDROME_USER = "" # TODO: load from env
|
||||
NAVIDROME_PASSWORD = "" # TODO: load from env
|
||||
|
||||
def _subsonic_params() -> dict:
|
||||
salt = secrets.token_hex(8)
|
||||
token = hashlib.md5((NAVIDROME_PASSWORD + salt).encode()).hexdigest()
|
||||
return {"u": NAVIDROME_USER, "t": token, "s": salt, "v": "1.16.1", "c": "muscriptor-mcp"}
|
||||
|
||||
async def search_track(query: str) -> list[dict]:
|
||||
"""Search Navidrome for tracks matching query. Returns list of {id, title, artist, duration}."""
|
||||
params = {**_subsonic_params(), "query": query, "songCount": 5}
|
||||
async with httpx.AsyncClient() as client:
|
||||
resp = await client.get(f"{NAVIDROME_URL}/rest/search3.view", params=params, timeout=15)
|
||||
resp.raise_for_status()
|
||||
data = resp.json()
|
||||
songs = data.get("subsonic-response", {}).get("searchResult3", {}).get("song", [])
|
||||
return [{"id": s["id"], "title": s["title"], "artist": s.get("artist", ""), "duration": s.get("duration", 0)} for s in songs]
|
||||
|
||||
async def download_track(song_id: str, output_dir: Path) -> Path:
|
||||
"""Download a track by ID. Returns local file path."""
|
||||
output_dir.mkdir(parents=True, exist_ok=True)
|
||||
params = {**_subsonic_params(), "id": song_id}
|
||||
async with httpx.AsyncClient() as client:
|
||||
resp = await client.get(f"{NAVIDROME_URL}/rest/download.view", params=params, timeout=120)
|
||||
resp.raise_for_status()
|
||||
# Navidrome sends the file with its original extension
|
||||
content_type = resp.headers.get("content-type", "audio/mpeg")
|
||||
ext = _ext_from_content_type(content_type)
|
||||
out_path = output_dir / f"{song_id}{ext}"
|
||||
out_path.write_bytes(resp.content)
|
||||
return out_path
|
||||
|
||||
def _ext_from_content_type(ct: str) -> str:
|
||||
mapping = {
|
||||
"audio/mpeg": ".mp3", "audio/flac": ".flac", "audio/ogg": ".ogg",
|
||||
"audio/wav": ".wav", "audio/x-wav": ".wav", "audio/mp4": ".m4a",
|
||||
}
|
||||
return mapping.get(ct.split(";")[0].strip(), ".mp3")
|
||||
```
|
||||
|
||||
**Verification:** Test search with `curl` against live Navidrome, confirm Subsonic auth works.
|
||||
|
||||
---
|
||||
|
||||
#### Task 4: Implement MuScriptor transcription engine
|
||||
|
||||
**Objective:** Wrap MuScriptor's `transcribe_to_midi` in an async wrapper.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/src/muscriptor_mcp/transcriber.py`
|
||||
|
||||
**Step 1:**
|
||||
|
||||
```python
|
||||
"""MuScriptor audio → MIDI transcription engine."""
|
||||
import asyncio
|
||||
from pathlib import Path
|
||||
from functools import lru_cache
|
||||
|
||||
@lru_cache(maxsize=1)
|
||||
def _load_model(size: str = "medium"):
|
||||
"""Load and cache the MuScriptor model (singleton per size)."""
|
||||
from muscriptor import TranscriptionModel
|
||||
return TranscriptionModel.load_model(size)
|
||||
|
||||
async def transcribe(audio_path: str | Path, output_path: str | Path | None = None,
|
||||
model_size: str = "medium", instruments: list[str] | None = None) -> Path:
|
||||
"""Transcribe audio file to MIDI. Returns path to .mid file."""
|
||||
audio_path = Path(audio_path)
|
||||
if output_path is None:
|
||||
output_path = audio_path.with_suffix(".mid")
|
||||
else:
|
||||
output_path = Path(output_path)
|
||||
|
||||
model = _load_model(model_size)
|
||||
|
||||
# MuScriptor's transcribe is CPU/GPU-bound, run in thread pool
|
||||
loop = asyncio.get_event_loop()
|
||||
midi_bytes = await loop.run_in_executor(None, lambda: model.transcribe_to_midi(str(audio_path)))
|
||||
output_path.write_bytes(midi_bytes)
|
||||
return output_path
|
||||
```
|
||||
|
||||
**Verification:** Run on a short WAV file to confirm model downloads + produces valid MIDI.
|
||||
|
||||
---
|
||||
|
||||
#### Task 5: Wire tools into MCP server
|
||||
|
||||
**Objective:** Connect all modules into the MCP server's tool handlers.
|
||||
|
||||
**Files:**
|
||||
- Modify: `scripts/muscriptor-mcp/src/muscriptor_mcp/server.py` — add `@app.call_tool()` handler
|
||||
|
||||
**Step 1:** Add the tool dispatch to `server.py`:
|
||||
|
||||
```python
|
||||
from muscriptor_mcp.maloja_client import get_top_tracks
|
||||
from muscriptor_mcp.navidrome_client import search_track, download_track
|
||||
from muscriptor_mcp.transcriber import transcribe
|
||||
|
||||
DEFAULT_DOWNLOAD_DIR = Path("/tmp/muscriptor")
|
||||
|
||||
@app.call_tool()
|
||||
async def call_tool(name: str, arguments: dict) -> list[TextContent]:
|
||||
if name == "transcribe_to_midi":
|
||||
audio_path = arguments["audio_path"]
|
||||
output_dir = arguments.get("output_dir")
|
||||
model_size = arguments.get("model_size", "medium")
|
||||
instruments = arguments.get("instruments")
|
||||
out = Path(output_dir) / Path(audio_path).with_suffix(".mid").name if output_dir else None
|
||||
result = await transcribe(audio_path, out, model_size, instruments)
|
||||
return [TextContent(type="text", text=f"MIDI saved to: {result}")]
|
||||
|
||||
elif name == "get_top_tracks":
|
||||
tracks = await get_top_tracks(arguments.get("count", 3), arguments.get("period", "week"))
|
||||
lines = [f"#{i+1}: {t.artist} — {t.title} ({t.play_count} plays)" for i, t in enumerate(tracks)]
|
||||
return [TextContent(type="text", text="\n".join(lines))]
|
||||
|
||||
elif name == "search_and_download":
|
||||
results = await search_track(arguments["query"])
|
||||
if not results:
|
||||
return [TextContent(type="text", text=f"No results for '{arguments['query']}'")]
|
||||
track = results[0]
|
||||
dl_dir = Path(arguments.get("download_dir", str(DEFAULT_DOWNLOAD_DIR)))
|
||||
path = await download_track(track["id"], dl_dir)
|
||||
return [TextContent(type="text", text=f"Downloaded: {track['artist']} — {track['title']} → {path}")]
|
||||
|
||||
elif name == "top_tracks_to_midi":
|
||||
count = arguments.get("count", 3)
|
||||
period = arguments.get("period", "week")
|
||||
model_size = arguments.get("model_size", "medium")
|
||||
tracks = await get_top_tracks(count, period)
|
||||
results = []
|
||||
dl_dir = DEFAULT_DOWNLOAD_DIR / "weekly"
|
||||
for t in tracks:
|
||||
query = f"{t.title} {t.artist}"
|
||||
search_results = await search_track(query)
|
||||
if not search_results:
|
||||
results.append(f"SKIP: {t.artist} — {t.title} (not found in Navidrome)")
|
||||
continue
|
||||
audio_path = await download_track(search_results[0]["id"], dl_dir)
|
||||
midi_path = await transcribe(audio_path, model_size=model_size)
|
||||
results.append(f"DONE: {t.artist} — {t.title} → {midi_path}")
|
||||
return [TextContent(type="text", text="\n".join(results))]
|
||||
|
||||
else:
|
||||
return [TextContent(type="text", text=f"Unknown tool: {name}")]
|
||||
```
|
||||
|
||||
**Verification:** `cd scripts/muscriptor-mcp && python -m muscriptor_mcp.server` starts without errors.
|
||||
|
||||
---
|
||||
|
||||
### Phase 2: Configuration & Secrets
|
||||
|
||||
#### Task 6: Environment-based secrets
|
||||
|
||||
**Objective:** Load API keys from environment variables, never hardcode.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/.env.example`
|
||||
- Modify: `scripts/muscriptor-mcp/src/muscriptor_mcp/maloja_client.py` — read from env
|
||||
- Modify: `scripts/muscriptor-mcp/src/muscriptor_mcp/navidrome_client.py` — read from env
|
||||
|
||||
**`.env.example`:**
|
||||
```bash
|
||||
MALOJA_URL=http://pf-maloja:42010
|
||||
MALOJA_API_KEY=your_maloja_api_key
|
||||
NAVIDROME_URL=http://pf-navidrome:4533
|
||||
NAVIDROME_USER=fabio
|
||||
NAVIDROME_PASSWORD=your_navidrome_password
|
||||
HF_TOKEN=your_huggingface_token
|
||||
```
|
||||
|
||||
**Verification:** Start server without .env → graceful error. Start with .env → tools work.
|
||||
|
||||
---
|
||||
|
||||
#### Task 7: Register MCP server in Aurelio config
|
||||
|
||||
**Objective:** Make the server discoverable by Aurelio agents.
|
||||
|
||||
**Files:**
|
||||
- Modify: `.aurelio/mcp_config.json` — add `muscriptor-mcp` entry
|
||||
|
||||
**Add entry:**
|
||||
```json
|
||||
"muscriptor-mcp": {
|
||||
"command": "python",
|
||||
"args": ["-m", "muscriptor_mcp.server"],
|
||||
"env": {
|
||||
"MALOJA_URL": "http://pf-maloja:42010",
|
||||
"NAVIDROME_URL": "http://pf-navidrome:4533"
|
||||
},
|
||||
"_disabled": false,
|
||||
"disabledTools": []
|
||||
}
|
||||
```
|
||||
|
||||
Note: secrets (API keys, passwords) go in the env file, not in mcp_config.json. The MCP server reads them from its own environment.
|
||||
|
||||
---
|
||||
|
||||
### Phase 3: Deployment on GPU Node
|
||||
|
||||
#### Task 8: Create systemd service for Gigabyte / Tomahawk MAX
|
||||
|
||||
**Objective:** Run the MCP server as a persistent service on the GPU node.
|
||||
|
||||
**Files:**
|
||||
- Create: `scripts/muscriptor-mcp/systemd/muscriptor-mcp.service`
|
||||
|
||||
**Systemd unit:**
|
||||
```ini
|
||||
[Unit]
|
||||
Description=MuScriptor MCP Server (audio → MIDI)
|
||||
After=network.target docker.service
|
||||
|
||||
[Service]
|
||||
Type=simple
|
||||
User=fabio
|
||||
WorkingDirectory=/home/fabio/muscriptor-mcp
|
||||
EnvironmentFile=/home/fabio/muscriptor-mcp/.env
|
||||
ExecStart=/home/fabio/muscriptor-mcp/.venv/bin/python -m muscriptor_mcp.server
|
||||
Restart=on-failure
|
||||
RestartSec=5
|
||||
# GPU access
|
||||
SupplementaryGroups=video render
|
||||
|
||||
[Install]
|
||||
WantedBy=multi-user.target
|
||||
```
|
||||
|
||||
**Deployment steps:**
|
||||
1. Clone repo to Gigabyte node (`192.168.0.104`)
|
||||
2. `cd scripts/muscriptor-mcp && python -m venv .venv && . .venv/bin/activate && pip install -e .`
|
||||
3. Create `.env` with real secrets
|
||||
4. Copy service file, enable, start
|
||||
5. First run will download ~1.2GB model weights (cached after)
|
||||
|
||||
---
|
||||
|
||||
#### Task 9: Expose via Cloudflare tunnel (optional)
|
||||
|
||||
**Objective:** Make MCP server accessible from Hermes gateway (CT 217) and laptop.
|
||||
|
||||
**Option A:** SSH tunnel from CT 217 to Gigabyte (preferred, no public exposure).
|
||||
**Option B:** Add Cloudflare tunnel route for `muscriptor-mcp.portugalfuturista.org`.
|
||||
|
||||
---
|
||||
|
||||
### Phase 4: Weekly Cron Job
|
||||
|
||||
#### Task 10: Hermes cron job for weekly top-3 MIDI conversion
|
||||
|
||||
**Objective:** Every Monday morning, auto-convert top 3 weekly tracks to MIDI.
|
||||
|
||||
**Using Hermes cron:**
|
||||
```
|
||||
Schedule: 0 9 * * 1 (every Monday at 09:00)
|
||||
Prompt: "Run the muscriptor-mcp top_tracks_to_midi tool with count=3, period=week, model_size=medium.
|
||||
Report which tracks were converted and where the MIDI files are saved."
|
||||
Deliver: telegram (or wherever the user wants)
|
||||
```
|
||||
|
||||
**Or via the MCP tool directly from any agent:**
|
||||
> "Hermes, convert my top 3 tracks this week to MIDI"
|
||||
|
||||
---
|
||||
|
||||
### Phase 5: Interactive Usage
|
||||
|
||||
#### Task 11: Document agent usage patterns
|
||||
|
||||
**Objective:** Show how users interact with the transcription tools.
|
||||
|
||||
**Usage patterns:**
|
||||
|
||||
1. **Ad-hoc transcription:**
|
||||
> "Hermes, here's Sun by Caribou. Convert it to MIDI for Ableton."
|
||||
→ Agent calls `search_and_download` then `transcribe_to_midi`
|
||||
|
||||
2. **Weekly auto-conversion:**
|
||||
→ Cron job runs `top_tracks_to_midi` automatically
|
||||
|
||||
3. **From a file path:**
|
||||
> "Transcribe /tmp/my_recording.wav to MIDI"
|
||||
→ Agent calls `transcribe_to_midi` directly
|
||||
|
||||
4. **With instrument restriction:**
|
||||
> "Transcribe just the piano from this track"
|
||||
→ Agent calls `transcribe_to_midi` with `instruments=["acoustic_piano"]`
|
||||
|
||||
5. **Batch conversion:**
|
||||
> "Convert all tracks from this album to MIDI"
|
||||
→ Agent searches Navidrome, loops `transcribe_to_midi`
|
||||
|
||||
---
|
||||
|
||||
## Files Summary
|
||||
|
||||
| Action | Path |
|
||||
|--------|------|
|
||||
| Create | `scripts/muscriptor-mcp/pyproject.toml` |
|
||||
| Create | `scripts/muscriptor-mcp/src/muscriptor_mcp/__init__.py` |
|
||||
| Create | `scripts/muscriptor-mcp/src/muscriptor_mcp/server.py` |
|
||||
| Create | `scripts/muscriptor-mcp/src/muscriptor_mcp/maloja_client.py` |
|
||||
| Create | `scripts/muscriptor-mcp/src/muscriptor_mcp/navidrome_client.py` |
|
||||
| Create | `scripts/muscriptor-mcp/src/muscriptor_mcp/transcriber.py` |
|
||||
| Create | `scripts/muscriptor-mcp/.env.example` |
|
||||
| Create | `scripts/muscriptor-mcp/systemd/muscriptor-mcp.service` |
|
||||
| Modify | `.aurelio/mcp_config.json` (add muscriptor-mcp entry) |
|
||||
|
||||
## Risks & Open Questions
|
||||
|
||||
1. **GPU VRAM:** MuScriptor medium (307M) should fit in RTX 3070 8GB. Large (1.4B) needs ~12GB — may OOM on 8GB cards. Use medium by default.
|
||||
2. **Audio format compatibility:** MuScriptor supports WAV natively. For MP3/FLAC/OGG, may need `ffmpeg` or `soundfile` for pre-conversion. Navidrome sends original format.
|
||||
3. **Maloja API shape:** Need to verify the exact API response structure — Maloja docs are sparse. Test with live instance first.
|
||||
4. **Navidrome auth:** Subsonic API uses token+salt auth. The multi-scrobbler config has credentials but they should go in .env, not committed.
|
||||
5. **Network path:** MCP server on Gigabyte needs to reach Maloja on CT 216 (192.168.0.126:42010) and Navidrome on Gigabyte Docker (pf-navidrome:4533). Verify LAN connectivity.
|
||||
6. **Model download:** First run downloads ~1.2GB from HuggingFace. Requires `HF_TOKEN` env var (free HF account).
|
||||
7. **Hermes cron limitation:** Cron jobs in TUI are local-only. For the weekly job to work reliably, either run it from the Hermes gateway on CT 217, or use a systemd timer on the GPU node.
|
||||
|
||||
## Dependencies to Install
|
||||
|
||||
```bash
|
||||
pip install muscriptor mcp[cli] httpx pydantic
|
||||
```
|
||||
|
||||
MuScriptor pulls in: `torch`, `torchaudio`, `transformers`, `safetensors`, `soundfile`.
|
||||
113
AGENTS.md
113
AGENTS.md
|
|
@ -97,6 +97,39 @@ small builtin set if absent). The Theia backend exposes the same data at
|
|||
is what populates the chat widget's provider picker and what
|
||||
`porta.portugalfuturista.org` re-serves at `/api/models/providers`.
|
||||
|
||||
## Mirror pipeline (upstream → self-hosted mirrors)
|
||||
|
||||
One-way replication of Savearth upstream sources into self-hosted mirrors,
|
||||
with RL feedback. Agents work on the mirrors; originals stay authoritative.
|
||||
|
||||
```
|
||||
GitHub (SavearthTech/*) ──► Forgejo (code.portugalfuturista.org)
|
||||
Jira Cloud ──► Plane (plane.portugalfuturista.org)
|
||||
Confluence Cloud ──► Outline (docs.portugalfuturista.org)
|
||||
```
|
||||
|
||||
| File | Role |
|
||||
|------|------|
|
||||
| `.aurelio/mirrors/sync-mirrors.yaml` | **Edit this** — sources, targets, auth refs |
|
||||
| `.aurelio/mirrors/dist/` | Generated catalog (do not edit) |
|
||||
| `scripts/sync-mirrors.py` | The engine: `--sync`, `--check`, `--write-in-place` |
|
||||
| `.forgejo/workflows/mirror-sync.yml` | CI: runs every 15 min |
|
||||
|
||||
Registry: `.aurelio/mirrors/sync-mirrors.yaml`. Tokens in env vars (never in
|
||||
the YAML) — see `.env.example` for the full list. Full guide:
|
||||
`.aurelio/mirrors/README.md`.
|
||||
|
||||
```bash
|
||||
python3 scripts/sync-mirrors.py --sync # all targets
|
||||
python3 scripts/sync-mirrors.py --sync forgejo # one target
|
||||
python3 scripts/sync-mirrors.py --write-in-place # regen dist after YAML edit
|
||||
python3 scripts/sync-mirrors.py --check # CI guard
|
||||
```
|
||||
|
||||
Sync outcomes emit RL reward signals to
|
||||
`.aurelio/brain/trajectory-rewards/mirror-sync.jsonl` — consumed by the
|
||||
GRPO/DPO trajectory training pipeline.
|
||||
|
||||
## CLI guides & service lenses
|
||||
|
||||
**Self-hosted first policy**: dev work targets our own services; public clouds
|
||||
|
|
@ -205,6 +238,86 @@ service disabled). Platform creds live in `/home/hermes/.hermes/.env` (`TELEGRAM
|
|||
`SIGNAL_*` (LAN-direct to .104:18088), `MATRIX_*`). State migrates via
|
||||
`hermes backup` → `hermes import`.
|
||||
|
||||
## Client onboarding (replicating into new projects)
|
||||
|
||||
Each Portugal Futurista client gets its own replica-omnisciente — a self-contained
|
||||
agent fleet brain with the same power set: `.aurelio/` (config cascade, brain,
|
||||
skills, providers, connectors, MCP, identity/heteronyms), `realms/`, `scripts/`,
|
||||
and CI/CD. The onboarding tool scaffolds this from the canonical template.
|
||||
|
||||
### Scaffold a new client
|
||||
|
||||
```bash
|
||||
# Interactive wizard (recommended for first-time use)
|
||||
python3 scripts/onboard-client.py --interactive
|
||||
|
||||
# CLI flags
|
||||
python3 scripts/onboard-client.py \
|
||||
--client-name "Acme Corp" \
|
||||
--git-url https://github.com/acme/replica-acme \
|
||||
--lead-engineer "Jane Doe" --lead-email "jane@acme.com" \
|
||||
--realm iot-firmware --realm-name "IoT Firmware" --realm-stack "ESP32, Rust" \
|
||||
--realm cloud-backend --realm-name "Cloud Backend" --realm-stack "Rust, Axum" \
|
||||
--output ~/clients/acme-corp
|
||||
|
||||
# JSON config file (reproducible, version-controlled)
|
||||
python3 scripts/onboard-client.py --config onboarding-acme.json
|
||||
|
||||
# Dry run (preview the file tree without writing)
|
||||
python3 scripts/onboard-client.py --client-name "Acme Corp" --dry-run
|
||||
```
|
||||
|
||||
### What gets scaffolded
|
||||
|
||||
| Component | Description |
|
||||
|-----------|-------------|
|
||||
| `AGENTS.md` + `README.md` + `.gitignore` + `.env.example` | Root project files with client identity |
|
||||
| `.aurelio/config.toml` | Three-tier config cascade root |
|
||||
| `.aurelio/brain/` + `memory/` + `plans/` + `sessions/` | Brain storage (empty, ready to fill) |
|
||||
| `.aurelio/identity/heteronyms.json` + `heteronimos/*.md` | Team members as heteronyms (voices) |
|
||||
| `.aurelio/providers/registry.yaml` | AI provider registry (OpenRouter, Gemini, Anthropic) |
|
||||
| `.aurelio/connectors/registry.yaml` | External integration catalog (git forge auto-detected) |
|
||||
| `.aurelio/skills/` | Full CLI skills tree copied from canonical replica |
|
||||
| `.aurelio/sync.py` | Brain sync utility (push/pull to Proxmox) |
|
||||
| `.aurelio/mcp_config.json` | MCP server fleet (empty, ready to configure) |
|
||||
| `.aurelio/chronicle/` | Event timeline + per-realm chronicle stubs |
|
||||
| `realms/<name>/` | Per-project: `AGENTS.md`, `.aurelio/config.toml`, `memory/index.md` |
|
||||
| `scripts/` | Onboarding CLI (self-replicating), agent importers, mirror generators |
|
||||
| `.forgejo/workflows/` + `.github/workflows/` | CI/CD for both forge types |
|
||||
|
||||
### Power toggles
|
||||
|
||||
Skip components with `--no-*` flags:
|
||||
```bash
|
||||
python3 scripts/onboard-client.py --client-name "Lite" --no-skills --no-connectors --no-ci
|
||||
```
|
||||
|
||||
### Managing realms after scaffolding
|
||||
|
||||
The scaffolded project includes its own `onboard-client.py` CLI:
|
||||
```bash
|
||||
cd <client-replica>
|
||||
python3 scripts/onboard-client.py add-realm my-project --name "My Project" --repo <url>
|
||||
python3 scripts/onboard-client.py list-realms
|
||||
python3 scripts/onboard-client.py info
|
||||
```
|
||||
|
||||
### Onboarding config format (JSON)
|
||||
|
||||
Save as `onboarding-<client>.json` for reproducible scaffolds:
|
||||
```json
|
||||
{
|
||||
"client_name": "Acme Corp",
|
||||
"description": "IoT infrastructure",
|
||||
"git_url": "https://github.com/acme/replica-acme",
|
||||
"lead_engineer_name": "Jane Doe",
|
||||
"lead_engineer_email": "jane@acme.com",
|
||||
"realms": [
|
||||
{"slug": "iot-fw", "name": "IoT Firmware", "stack": "ESP32, Rust"}
|
||||
]
|
||||
}
|
||||
```
|
||||
|
||||
## Gotchas
|
||||
|
||||
- `extensions/aurelio-vscode/` and `aurelio-theia/` are git submodules — run `git submodule update --init` after clone; edit them in their own repos (`~/portugalfuturista/aurelio-vscode`, `~/portugalfuturista/aurelio-theia`) and push to Forgejo.
|
||||
|
|
|
|||
392
scripts/brain-to-gbrain.py
Normal file
392
scripts/brain-to-gbrain.py
Normal file
|
|
@ -0,0 +1,392 @@
|
|||
#!/usr/bin/env python3
|
||||
"""
|
||||
Bridge: Réplica brain sessions → gbrain conversation pages.
|
||||
|
||||
Reads .aurelio/brain/session-*/ directories and pushes each session as a
|
||||
gbraidge 'conversation' page via the gbrain CLI.
|
||||
|
||||
Output per session:
|
||||
slug: conversations/<source>/<session-id>
|
||||
body: YAML frontmatter + formatted transcript
|
||||
|
||||
Usage:
|
||||
python3 scripts/brain-to-gbrain.py [--dry-run] [--source NAME]...
|
||||
[--replica-root PATH]
|
||||
[--ssh HOST] [--gbrain-bin PATH]
|
||||
[--summary] [--limit N]
|
||||
|
||||
Modes:
|
||||
--ssh USER@HOST Push via SSH to remote gbrain (CT 208).
|
||||
Default: root@192.168.0.38 (Proxmox → pct 208)
|
||||
(local) Use local gbrain CLI (must be initialized locally).
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import argparse
|
||||
import json
|
||||
import subprocess
|
||||
import sys
|
||||
import tempfile
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
_here = Path(__file__).resolve()
|
||||
for _candidate in (_here.parent, *_here.parents):
|
||||
if (_candidate / ".aurelio").is_dir():
|
||||
REPLICA_ROOT = _candidate
|
||||
break
|
||||
else:
|
||||
REPLICA_ROOT = _here.parents[2]
|
||||
|
||||
BRAIN_DIR = REPLICA_ROOT / ".aurelio" / "brain"
|
||||
|
||||
VALID_SOURCES = {
|
||||
"kimi", "hermes", "claude-code", "antigravity", "qwen-code",
|
||||
"mimocode", "pi", "opencode",
|
||||
}
|
||||
|
||||
DEFAULT_SSH = "root@192.168.0.38"
|
||||
REMOTE_GBRAIN_BIN = "/opt/pf-services-208/aurelio-gbrain/bin/gbrain"
|
||||
REMOTE_CT = "208"
|
||||
|
||||
|
||||
def _load_json(path: Path) -> Any:
|
||||
try:
|
||||
return json.loads(path.read_text(encoding="utf-8"))
|
||||
except (OSError, json.JSONDecodeError):
|
||||
return None
|
||||
|
||||
|
||||
def _load_chat_history(session_dir: Path) -> dict | None:
|
||||
ch = session_dir / ".system_generated" / "chat_history.json"
|
||||
return _load_json(ch)
|
||||
|
||||
|
||||
def _load_session_meta(session_dir: Path) -> dict | None:
|
||||
sj = session_dir / "session.jsonl"
|
||||
if not sj.exists():
|
||||
return None
|
||||
lines = sj.read_text(encoding="utf-8").strip().split("\n")
|
||||
meta = {}
|
||||
for line in lines:
|
||||
try:
|
||||
rec = json.loads(line)
|
||||
except json.JSONDecodeError:
|
||||
continue
|
||||
if rec.get("type") == "session_start":
|
||||
meta.update(rec.get("data", {}))
|
||||
meta["created"] = rec.get("timestamp")
|
||||
elif rec.get("type") == "session_end":
|
||||
meta["totalEvents"] = rec.get("data", {}).get("totalEvents", 0)
|
||||
return meta or None
|
||||
|
||||
|
||||
def _format_transcript(chat_history: dict) -> str:
|
||||
messages = chat_history.get("messages", [])
|
||||
lines = []
|
||||
for msg in messages:
|
||||
role = msg.get("role", "unknown")
|
||||
content = msg.get("content", "")
|
||||
if not content:
|
||||
continue
|
||||
model = msg.get("model") or msg.get("modelType", "")
|
||||
if role == "user":
|
||||
lines.append(f"## User\n\n{content}\n")
|
||||
elif role == "assistant":
|
||||
header = f"## Assistant"
|
||||
if model:
|
||||
header += f" ({model})"
|
||||
lines.append(f"{header}\n\n{content}\n")
|
||||
elif role in ("tool_call", "tool_result"):
|
||||
lines.append(f"### {role}\n\n{content[:800]}\n")
|
||||
elif role == "system":
|
||||
lines.append(f"### System\n\n{content[:500]}\n")
|
||||
return "\n".join(lines)
|
||||
|
||||
|
||||
def _make_page_content(session_dir: Path, chat_history: dict, meta: dict | None) -> str:
|
||||
source = chat_history.get("source", "unknown")
|
||||
title = chat_history.get("title", f"{source} session")
|
||||
created = chat_history.get("created", "")
|
||||
modified = chat_history.get("modified", "")
|
||||
workspace = chat_history.get("workspacePath", "")
|
||||
session_id = chat_history.get("id", session_dir.name)
|
||||
source_locator = chat_history.get("sourceLocator", "")
|
||||
model_type = chat_history.get("modelType", source)
|
||||
messages = chat_history.get("messages", [])
|
||||
total_events = meta.get("totalEvents", len(messages)) if meta else len(messages)
|
||||
|
||||
tags = [source, "agent-session"]
|
||||
if workspace:
|
||||
tags.append(Path(workspace).name if workspace else "")
|
||||
tags = [t for t in tags if t]
|
||||
|
||||
frontmatter_lines = [
|
||||
"---",
|
||||
"type: conversation",
|
||||
f'title: "{title.replace(chr(34), chr(39))}"',
|
||||
f"date: {created[:10] if created else 'unknown'}",
|
||||
f"source: {source}",
|
||||
f"model: {model_type}",
|
||||
f"tags: [{', '.join(tags)}]",
|
||||
f"workspace: {workspace or 'unknown'}",
|
||||
f"brain_session_id: {session_id}",
|
||||
f"message_count: {len(messages)}",
|
||||
f"total_events: {total_events}",
|
||||
]
|
||||
if source_locator:
|
||||
frontmatter_lines.append(f"source_locator: \"{source_locator}\"")
|
||||
if modified:
|
||||
frontmatter_lines.append(f"modified: {modified}")
|
||||
frontmatter_lines.append("---")
|
||||
frontmatter_lines.append("")
|
||||
|
||||
body_lines = [
|
||||
f"# {title}",
|
||||
"",
|
||||
f"> Source: **{source}** | Model: **{model_type}** | Messages: **{len(messages)}**",
|
||||
]
|
||||
if workspace:
|
||||
body_lines.append(f"> Workspace: `{workspace}`")
|
||||
body_lines.append("")
|
||||
body_lines.append("## Transcript")
|
||||
body_lines.append("")
|
||||
body_lines.append(_format_transcript(chat_history))
|
||||
|
||||
return "\n".join(frontmatter_lines + body_lines)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Remote SSH mode: stage files locally, then import on CT 208 via --no-embed
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
def _ssh_gbrain_import(ssh_host: str, staging_dir: Path, dry_run: bool) -> dict:
|
||||
"""SCP staging dir to CT 208, run gbrain import --no-embed."""
|
||||
if dry_run:
|
||||
count = len(list(staging_dir.glob("*.md")))
|
||||
return {"ok": True, "imported": count, "dry_run": True}
|
||||
|
||||
try:
|
||||
# Tar + scp for efficiency — extract to /tmp so files land at /tmp/<staging_dir_name>/
|
||||
tar_cmd = f"tar czf - -C {staging_dir.parent} {staging_dir.name}"
|
||||
ssh_cmd = f"ssh -o ConnectTimeout=10 {ssh_host} pct exec {REMOTE_CT} -- tar xzf - -C /tmp"
|
||||
tar_proc = subprocess.Popen(tar_cmd, shell=True, stdout=subprocess.PIPE)
|
||||
ssh_proc = subprocess.Popen(ssh_cmd, shell=True, stdin=tar_proc.stdout, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
tar_proc.stdout.close()
|
||||
tar_proc.wait()
|
||||
ssh_proc.wait()
|
||||
|
||||
if tar_proc.returncode != 0 or ssh_proc.returncode != 0:
|
||||
return {"ok": False, "error": f"tar/scp failed: {ssh_proc.stderr.decode()[:200]}"}
|
||||
|
||||
remote_import_path = f"/tmp/{staging_dir.name}"
|
||||
|
||||
# Run gbrain import --no-embed
|
||||
import_cmd = (
|
||||
f"pct exec {REMOTE_CT} -- {REMOTE_GBRAIN_BIN} "
|
||||
f"import {remote_import_path} --no-embed"
|
||||
)
|
||||
r = subprocess.run(
|
||||
["ssh", "-o", "ConnectTimeout=10", ssh_host, import_cmd],
|
||||
capture_output=True, text=True, timeout=600,
|
||||
)
|
||||
|
||||
# Cleanup remote
|
||||
subprocess.run(
|
||||
["ssh", "-o", "ConnectTimeout=5", ssh_host,
|
||||
f"pct exec {REMOTE_CT} -- rm -rf {remote_import_path}"],
|
||||
capture_output=True, timeout=15,
|
||||
)
|
||||
|
||||
if r.returncode == 0:
|
||||
output = r.stdout + r.stderr
|
||||
imported = 0
|
||||
for line in output.split("\n"):
|
||||
if "imported" in line and "pages" in line:
|
||||
parts = line.split()
|
||||
for i, p in enumerate(parts):
|
||||
if p == "imported" and i > 0:
|
||||
try:
|
||||
imported = int(parts[i - 1])
|
||||
except ValueError:
|
||||
pass
|
||||
return {"ok": True, "imported": imported, "output": output.strip()[:500]}
|
||||
# Even with non-zero exit, check if pages were imported (warnings cause exit 1)
|
||||
output = (r.stdout or "") + (r.stderr or "")
|
||||
imported = 0
|
||||
for line in output.split("\n"):
|
||||
if "imported" in line and "pages" in line:
|
||||
parts = line.split()
|
||||
for i, p in enumerate(parts):
|
||||
if p == "imported" and i > 0:
|
||||
try:
|
||||
imported = int(parts[i - 1])
|
||||
except ValueError:
|
||||
pass
|
||||
if imported > 0:
|
||||
return {"ok": True, "imported": imported, "output": output.strip()[:500]}
|
||||
return {"ok": False, "error": output.strip()[:500]}
|
||||
|
||||
except (FileNotFoundError, subprocess.TimeoutExpired) as e:
|
||||
return {"ok": False, "error": str(e)}
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Local mode: put pages via local gbrain CLI
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
def _local_gbrain_put(gbrain_bin: str | None, slug: str, content: str, dry_run: bool) -> dict:
|
||||
if dry_run:
|
||||
return {"ok": True, "slug": slug, "dry_run": True, "content_length": len(content)}
|
||||
|
||||
cmd = [gbrain_bin or "gbrain", "put", slug]
|
||||
try:
|
||||
r = subprocess.run(cmd, input=content, capture_output=True, text=True, timeout=120)
|
||||
if r.returncode == 0:
|
||||
return {"ok": True, "slug": slug}
|
||||
return {"ok": False, "slug": slug, "error": r.stderr.strip() or r.stdout.strip()}
|
||||
except (FileNotFoundError, subprocess.TimeoutExpired) as e:
|
||||
return {"ok": False, "slug": slug, "error": str(e)}
|
||||
|
||||
|
||||
def discover_sessions(brain_dir: Path, source_filter: set[str] | None = None) -> list[Path]:
|
||||
sessions = []
|
||||
for d in sorted(brain_dir.iterdir()):
|
||||
if not d.is_dir() or not d.name.startswith("session-"):
|
||||
continue
|
||||
parts = d.name.split("-", 2)
|
||||
if len(parts) < 2:
|
||||
continue
|
||||
source = parts[1]
|
||||
if source_filter and source not in source_filter:
|
||||
continue
|
||||
if source not in VALID_SOURCES:
|
||||
continue
|
||||
sessions.append(d)
|
||||
return sessions
|
||||
|
||||
|
||||
def run(args: argparse.Namespace) -> dict[str, Any]:
|
||||
replica_root = Path(args.replica_root).resolve() if args.replica_root else REPLICA_ROOT
|
||||
brain_dir = replica_root / ".aurelio" / "brain"
|
||||
|
||||
source_filter = set(args.source) if args.source else None
|
||||
sessions = discover_sessions(brain_dir, source_filter)
|
||||
|
||||
if args.limit:
|
||||
sessions = sessions[:args.limit]
|
||||
|
||||
ssh_host = args.ssh
|
||||
use_remote = bool(ssh_host)
|
||||
|
||||
summary: dict[str, Any] = {
|
||||
"replica_root": str(replica_root),
|
||||
"brain_dir": str(brain_dir),
|
||||
"dry_run": args.dry_run,
|
||||
"mode": "remote-ssh" if use_remote else "local",
|
||||
"remote": ssh_host or None,
|
||||
"total_sessions": len(sessions),
|
||||
"pushed": 0,
|
||||
"failed": 0,
|
||||
"errors": [],
|
||||
}
|
||||
|
||||
if use_remote:
|
||||
# Stage all files locally, then bulk import
|
||||
staging = replica_root / ".aurelio" / "gbraidge-staging"
|
||||
staging.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
for sess_dir in sessions:
|
||||
chat_history = _load_chat_history(sess_dir)
|
||||
if not chat_history:
|
||||
continue
|
||||
meta = _load_session_meta(sess_dir)
|
||||
source = chat_history.get("source", "unknown")
|
||||
session_id = chat_history.get("id", sess_dir.name)
|
||||
slug = f"conversations/{source}/{session_id.removeprefix('session-')}"
|
||||
content = _make_page_content(sess_dir, chat_history, meta)
|
||||
|
||||
out = staging / f"{slug.replace('/', '_')}.md"
|
||||
out.parent.mkdir(parents=True, exist_ok=True)
|
||||
if not args.dry_run:
|
||||
out.write_text(content, encoding="utf-8")
|
||||
summary["pushed"] += 1
|
||||
|
||||
# Bulk import
|
||||
if not args.dry_run:
|
||||
print(f"Staged {summary['pushed']} files. Importing to gbrain...", file=sys.stderr)
|
||||
result = _ssh_gbrain_import(ssh_host, staging, args.dry_run)
|
||||
summary["import_result"] = result
|
||||
if not result.get("ok"):
|
||||
summary["failed"] = summary["pushed"]
|
||||
summary["errors"].append({"error": result.get("error", "unknown")})
|
||||
else:
|
||||
summary["imported"] = result.get("imported", 0)
|
||||
|
||||
# Cleanup staging
|
||||
if not args.dry_run:
|
||||
import shutil
|
||||
shutil.rmtree(staging, ignore_errors=True)
|
||||
else:
|
||||
for sess_dir in sessions:
|
||||
chat_history = _load_chat_history(sess_dir)
|
||||
if not chat_history:
|
||||
continue
|
||||
meta = _load_session_meta(sess_dir)
|
||||
source = chat_history.get("source", "unknown")
|
||||
session_id = chat_history.get("id", sess_dir.name)
|
||||
slug = f"conversations/{source}/{session_id.removeprefix('session-')}"
|
||||
content = _make_page_content(sess_dir, chat_history, meta)
|
||||
|
||||
result = _local_gbrain_put(args.gbrain_bin, slug, content, args.dry_run)
|
||||
if result["ok"]:
|
||||
summary["pushed"] += 1
|
||||
if args.verbose:
|
||||
print(f" OK {slug}")
|
||||
else:
|
||||
summary["failed"] += 1
|
||||
err = result.get("error", "unknown")
|
||||
summary["errors"].append({"slug": slug, "error": err})
|
||||
if args.verbose:
|
||||
print(f" FAIL {slug}: {err}", file=sys.stderr)
|
||||
|
||||
return summary
|
||||
|
||||
|
||||
def main() -> int:
|
||||
ap = argparse.ArgumentParser(description="Bridge brain sessions → gbrain conversation pages.")
|
||||
ap.add_argument("--dry-run", action="store_true", help="Report what would be pushed without touching gbrain.")
|
||||
ap.add_argument("--source", action="append", help="Limit to specific sources (repeatable).")
|
||||
ap.add_argument("--replica-root", default=None)
|
||||
ap.add_argument("--ssh", default=None,
|
||||
help=f"SSH target for remote gbrain (default: {DEFAULT_SSH}). "
|
||||
"Set to empty string for local mode.")
|
||||
ap.add_argument("--gbrain-bin", default=None, help="Path to local gbrain binary.")
|
||||
ap.add_argument("--limit", type=int, default=None, help="Max sessions to process.")
|
||||
ap.add_argument("--summary", action="store_true", help="Print JSON summary.")
|
||||
ap.add_argument("--verbose", "-v", action="store_true")
|
||||
args = ap.parse_args()
|
||||
|
||||
if args.ssh is None:
|
||||
args.ssh = DEFAULT_SSH
|
||||
|
||||
summary = run(args)
|
||||
|
||||
if args.summary:
|
||||
print(json.dumps(summary, indent=2, ensure_ascii=False))
|
||||
else:
|
||||
print(f"mode: {summary['mode']} remote: {summary['remote'] or 'local'}")
|
||||
print(f"sessions: {summary['total_sessions']} pushed: {summary['pushed']} failed: {summary['failed']}")
|
||||
if summary.get("imported"):
|
||||
print(f"imported: {summary['imported']}")
|
||||
if args.dry_run:
|
||||
print("(dry-run: nothing pushed)")
|
||||
for err in summary.get("errors", []):
|
||||
print(f" FAIL: {err.get('slug', 'batch')}: {err.get('error', 'unknown')}", file=sys.stderr)
|
||||
|
||||
return 1 if summary.get("failed", 0) > 0 else 0
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
raise SystemExit(main())
|
||||
318
scripts/vault-sync.py
Normal file
318
scripts/vault-sync.py
Normal file
|
|
@ -0,0 +1,318 @@
|
|||
#!/usr/bin/env python3
|
||||
"""
|
||||
vault-sync — Obsidian vault → GBrain ingestion daemon (MCP).
|
||||
|
||||
Watches a directory (synced from phone via Nextcloud/Remotely Save),
|
||||
ingests markdown notes into GBrain via its MCP HTTP endpoint.
|
||||
|
||||
Runs as a systemd service on CT223 (pf-forja-do-conhecimento).
|
||||
|
||||
Config: /etc/vault-sync.json
|
||||
{
|
||||
"vault_dir": "/mnt/usb-pool/obsidian-vault",
|
||||
"gbrain_url": "http://127.0.0.1:18001/mcp",
|
||||
"gbrain_token": "gbrain_at_...",
|
||||
"poll_interval": 300,
|
||||
"min_file_age": 5,
|
||||
"state_db": "/var/lib/vault-sync/state.db"
|
||||
}
|
||||
"""
|
||||
|
||||
import hashlib
|
||||
import json
|
||||
import os
|
||||
import re
|
||||
import signal
|
||||
import sqlite3
|
||||
import sys
|
||||
import time
|
||||
import urllib.request
|
||||
import urllib.error
|
||||
from datetime import datetime, timezone
|
||||
from pathlib import Path
|
||||
|
||||
DEFAULT_CONFIG = {
|
||||
"vault_dir": "/mnt/usb-pool/obsidian-vault",
|
||||
"gbrain_url": "http://127.0.0.1:18001/mcp",
|
||||
"gbrain_token": "",
|
||||
"poll_interval": 300,
|
||||
"min_file_age": 5,
|
||||
"state_db": "/var/lib/vault-sync/state.db",
|
||||
"source_slug": "obsidian-vault",
|
||||
}
|
||||
|
||||
WIKILINK_RE = re.compile(r"\[\[([^\]|]+)(?:\|[^\]]+)?\]\]")
|
||||
FRONTMATTER_RE = re.compile(r"^---\n(.*?)\n---\n", re.DOTALL)
|
||||
TAG_LINE_RE = re.compile(r"^tags?:\s*(.+)$", re.MULTILINE)
|
||||
INLINE_TAG_RE = re.compile(r"(?:^|\s)#([a-zA-Z][a-zA-Z0-9_/\-]*)")
|
||||
|
||||
|
||||
def mcp_call(url, token, tool, args=None):
|
||||
"""Call a GBrain MCP tool via HTTP JSON-RPC."""
|
||||
params = {"name": tool, "arguments": args or {}}
|
||||
data = json.dumps({"jsonrpc": "2.0", "method": "tools/call", "params": params, "id": 1})
|
||||
req = urllib.request.Request(
|
||||
url,
|
||||
data=data.encode(),
|
||||
headers={
|
||||
"Content-Type": "application/json",
|
||||
"Accept": "application/json, text/event-stream",
|
||||
"Authorization": f"Bearer {token}",
|
||||
},
|
||||
)
|
||||
with urllib.request.urlopen(req, timeout=30) as resp:
|
||||
raw = resp.read().decode()
|
||||
for line in raw.splitlines():
|
||||
if line.startswith("data: "):
|
||||
return json.loads(line[6:])
|
||||
return json.loads(raw)
|
||||
|
||||
|
||||
def slugify(text):
|
||||
"""Convert a note title to a GBrain-safe slug."""
|
||||
s = text.lower().strip()
|
||||
s = re.sub(r"[^a-z0-9\s/-]", "", s)
|
||||
s = re.sub(r"[\s/]+", "-", s)
|
||||
s = re.sub(r"-+", "-", s).strip("-")
|
||||
return s[:80] or "untitled"
|
||||
|
||||
|
||||
class VaultSync:
|
||||
def __init__(self, config_path="/etc/vault-sync.json"):
|
||||
self.config = dict(DEFAULT_CONFIG)
|
||||
try:
|
||||
with open(config_path) as f:
|
||||
self.config.update(json.load(f))
|
||||
except (OSError, json.JSONDecodeError):
|
||||
pass
|
||||
|
||||
self.vault_dir = Path(self.config["vault_dir"])
|
||||
self.gbrain_url = self.config["gbrain_url"].rstrip("/")
|
||||
self.gbrain_token = self.config["gbrain_token"]
|
||||
self.poll_interval = self.config["poll_interval"]
|
||||
self.min_file_age = self.config["min_file_age"]
|
||||
self.source_slug = self.config["source_slug"]
|
||||
|
||||
os.makedirs(os.path.dirname(self.config["state_db"]), exist_ok=True)
|
||||
self.db = sqlite3.connect(self.config["state_db"])
|
||||
self.db.execute("""
|
||||
CREATE TABLE IF NOT EXISTS ingested (
|
||||
path TEXT PRIMARY KEY,
|
||||
mtime REAL,
|
||||
content_hash TEXT,
|
||||
gbrain_slug TEXT,
|
||||
ingested_at TEXT
|
||||
)
|
||||
""")
|
||||
self.db.commit()
|
||||
|
||||
self.running = True
|
||||
signal.signal(signal.SIGTERM, self._shutdown)
|
||||
signal.signal(signal.SIGINT, self._shutdown)
|
||||
|
||||
def _shutdown(self, *_):
|
||||
print("Shutting down...", flush=True)
|
||||
self.running = False
|
||||
|
||||
def run(self):
|
||||
print(f"vault-sync: {self.vault_dir} → {self.gbrain_url}", flush=True)
|
||||
print(f" poll: {self.poll_interval}s, min_age: {self.min_file_age}s", flush=True)
|
||||
|
||||
while self.running:
|
||||
try:
|
||||
self.sync_once()
|
||||
except Exception as e:
|
||||
print(f"Sync error: {e}", file=sys.stderr, flush=True)
|
||||
for _ in range(int(self.poll_interval)):
|
||||
if not self.running:
|
||||
break
|
||||
time.sleep(1)
|
||||
|
||||
self.db.close()
|
||||
print("vault-sync stopped.", flush=True)
|
||||
|
||||
def sync_once(self):
|
||||
if not self.vault_dir.exists():
|
||||
print(f" vault dir not found: {self.vault_dir}", flush=True)
|
||||
return
|
||||
|
||||
notes = sorted(self.vault_dir.rglob("*.md"))
|
||||
total = len(notes)
|
||||
new = changed = skipped = errors = 0
|
||||
|
||||
for note_path in notes:
|
||||
if not self.running:
|
||||
break
|
||||
rel = note_path.relative_to(self.vault_dir)
|
||||
stat = note_path.stat()
|
||||
age = time.time() - stat.st_mtime
|
||||
|
||||
if age < self.min_file_age:
|
||||
skipped += 1
|
||||
continue
|
||||
|
||||
row = self.db.execute(
|
||||
"SELECT mtime FROM ingested WHERE path = ?", (str(rel),)
|
||||
).fetchone()
|
||||
if row and row[0] == stat.st_mtime:
|
||||
skipped += 1
|
||||
continue
|
||||
|
||||
try:
|
||||
result = self.ingest_note(note_path, rel)
|
||||
if result == "new":
|
||||
new += 1
|
||||
elif result == "updated":
|
||||
changed += 1
|
||||
else:
|
||||
skipped += 1
|
||||
except Exception as e:
|
||||
errors += 1
|
||||
print(f" ERROR {rel}: {e}", file=sys.stderr, flush=True)
|
||||
|
||||
if new or changed or errors:
|
||||
print(
|
||||
f" {total} notes: {new} new, {changed} updated, "
|
||||
f"{skipped} skipped, {errors} errors",
|
||||
flush=True,
|
||||
)
|
||||
|
||||
def ingest_note(self, note_path, rel):
|
||||
content = note_path.read_text(encoding="utf-8", errors="replace")
|
||||
if not content.strip():
|
||||
return "skip"
|
||||
|
||||
content_hash = hashlib.sha1(content.encode()).hexdigest()[:16]
|
||||
stat = note_path.stat()
|
||||
|
||||
# Parse frontmatter
|
||||
frontmatter = {}
|
||||
body = content
|
||||
fm = FRONTMATTER_RE.match(content)
|
||||
if fm:
|
||||
body = content[fm.end():]
|
||||
for line in fm.group(1).splitlines():
|
||||
if ":" in line:
|
||||
k, v = line.split(":", 1)
|
||||
frontmatter[k.strip().lower()] = v.strip()
|
||||
|
||||
# Title: frontmatter title > first heading > filename
|
||||
title = frontmatter.get("title", "")
|
||||
if not title:
|
||||
heading = re.search(r"^#\s+(.+)$", body, re.MULTILINE)
|
||||
title = heading.group(1).strip() if heading else note_path.stem
|
||||
|
||||
# Wikilinks
|
||||
wikilinks = list(set(WIKILINK_RE.findall(body)))
|
||||
|
||||
# Tags: frontmatter tags + inline #tags
|
||||
tags = set()
|
||||
tag_match = TAG_LINE_RE.search(content)
|
||||
if tag_match:
|
||||
raw = tag_match.group(1)
|
||||
# Handle YAML arrays: [tag1, tag2] or plain: tag1, tag2
|
||||
if raw.startswith("[") and raw.endswith("]"):
|
||||
raw = raw[1:-1]
|
||||
for t in raw.split(","):
|
||||
t = t.strip().strip("'\"[]").lstrip("#")
|
||||
if t:
|
||||
tags.add(t)
|
||||
for t in INLINE_TAG_RE.findall(body):
|
||||
tags.add(t)
|
||||
tags = sorted(tags - {""})
|
||||
|
||||
# Slug for GBrain page
|
||||
slug = slugify(str(rel).replace(".md", ""))
|
||||
|
||||
# Check if this is new or update
|
||||
existing = self.db.execute(
|
||||
"SELECT gbrain_slug FROM ingested WHERE path = ?", (str(rel),)
|
||||
).fetchone()
|
||||
is_update = existing is not None
|
||||
|
||||
# Build GBrain page content with frontmatter
|
||||
gbrain_frontmatter = {
|
||||
"type": "note",
|
||||
"source": self.source_slug,
|
||||
"title": title,
|
||||
"original_path": str(rel),
|
||||
"content_hash": content_hash,
|
||||
"file_mtime": stat.st_mtime,
|
||||
"synced_at": datetime.now(timezone.utc).isoformat(),
|
||||
**{k: v for k, v in frontmatter.items() if k not in ("title",)},
|
||||
}
|
||||
fm_lines = ["---"]
|
||||
for k, v in gbrain_frontmatter.items():
|
||||
if isinstance(v, str) and (":" in v or '"' in v):
|
||||
v = json.dumps(v)
|
||||
elif isinstance(v, (int, float)):
|
||||
v = str(v)
|
||||
elif isinstance(v, list):
|
||||
v = json.dumps(v)
|
||||
fm_lines.append(f"{k}: {v}")
|
||||
fm_lines.append("---\n")
|
||||
|
||||
page_content = "\n".join(fm_lines) + body
|
||||
|
||||
# PUT page to GBrain
|
||||
try:
|
||||
result = mcp_call(self.gbrain_url, self.gbrain_token, "put_page", {
|
||||
"slug": slug,
|
||||
"content": page_content,
|
||||
})
|
||||
gbrain_slug = slug
|
||||
except Exception as e:
|
||||
raise RuntimeError(f"put_page failed: {e}")
|
||||
|
||||
# Add tags
|
||||
for tag in tags[:20]:
|
||||
try:
|
||||
mcp_call(self.gbrain_url, self.gbrain_token, "add_tag", {
|
||||
"slug": gbrain_slug,
|
||||
"tag": tag,
|
||||
})
|
||||
except Exception:
|
||||
pass # tag errors are non-fatal
|
||||
|
||||
# Add wikilinks as graph links
|
||||
for link_target in wikilinks[:50]:
|
||||
target_slug = slugify(link_target)
|
||||
try:
|
||||
mcp_call(self.gbrain_url, self.gbrain_token, "add_link", {
|
||||
"from": gbrain_slug,
|
||||
"to": target_slug,
|
||||
"link_type": "wikilink",
|
||||
})
|
||||
except Exception:
|
||||
pass # link errors non-fatal (target may not exist yet)
|
||||
|
||||
# Log the ingest
|
||||
try:
|
||||
mcp_call(self.gbrain_url, self.gbrain_token, "log_ingest", {
|
||||
"source": self.source_slug,
|
||||
"external_id": str(rel),
|
||||
"action": "update" if is_update else "create",
|
||||
"title": title,
|
||||
})
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
now = datetime.now(timezone.utc).isoformat()
|
||||
self.db.execute(
|
||||
"""INSERT OR REPLACE INTO ingested (path, mtime, content_hash, gbrain_slug, ingested_at)
|
||||
VALUES (?, ?, ?, ?, ?)""",
|
||||
(str(rel), stat.st_mtime, content_hash, gbrain_slug, now),
|
||||
)
|
||||
self.db.commit()
|
||||
|
||||
return "updated" if is_update else "new"
|
||||
|
||||
|
||||
def main():
|
||||
config_path = os.environ.get("VAULT_SYNC_CONFIG", "/etc/vault-sync.json")
|
||||
sync = VaultSync(config_path)
|
||||
sync.run()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Loading…
Reference in a new issue