2 commits

Author	SHA1	Message	Date
Raphael Cautus (Maestro)	c96b6631ce	feat(scripts): agent importer — new adapters, RL trajectory output ... Extends sync-agents-to-brain.py with: - cursor, zed, continue, aider, cline adapters (11 total now) - RL trajectory extraction (state/action/observation triples) - Improved content flattening for OpenAI/Anthropic multipart content - Engine: stable per-message timestamps, session_memory.md rendering Co-authored-by: Álvaro de Campos <campos@portugalfuturista.org>	2026-07-31 00:08:26 +01:00
fabiorafaelcoutada	c2cdc4f939	feat(brain): unified agent -> brain importer across coding agents ... scripts/sync-agents-to-brain.py fans every coding-agent's local artifacts into the brain as session-<source>-<id>/ (session.jsonl, session_memory.md, summary.md, metadata.json), reusing the Kimi layout contract. - scripts/agent_importers/engine.py shared normalizer + content-addressed writer (idempotent: re-run writes 0 files) - scripts/agent_importers/adapters.py discover_* for kimi, hermes, claude-code (+ subagents), antigravity (SQLite), qwen-code, mimocode - scripts/sync-kimi-to-brain.py now a thin wrapper delegating to --source kimi (existing calls keep working) Flags: --dry-run, --skip-active, --source, --replica-root, --summary. Run before .aurelio/sync.py --push to broadcast to the fleet.	2026-07-10 22:39:30 +01:00