replica-omnisciente/research/pt-futurista-audit/findings/F2.md
2026-07-12 20:53:08 +01:00

1.9 KiB
Raw Blame History

F2: Security posture across both repos

Findings

[1] Lidarr API key hardcoded in 5 Python scripts

  • quote: "api_key = "24077d03b9a742a786dd799567659cb8""
  • url: local (nervura-electrica/rename_lidarr.py:7)
  • source_type: primary
  • published: 2026
  • confidence: high

[2] Telegram bot token hardcoded in watchdog config

  • quote: "bot_token: "8992450776:AAHx5xta6BDL3OR7WnWsukWSasCZh3U288E""
  • url: local (nervura-electrica/infrastructure/proxmox/watchdog/config.yml:13)
  • source_type: primary
  • published: 2026
  • confidence: high

[3] Session API key leaked in tracked YAML

  • quote: "api_key: "bff75943e17f48fb060f1c6321c28b513cad7e8b876c00b11b16e2d067edcdc3""
  • url: local (replica-omnisciente/.aurelio/sessions/2026-04-30_antigravity-cdp-integration/session.yaml:93)
  • source_type: primary
  • published: 2026
  • confidence: high

[4] 43% of MCP servers vulnerable to command injection

[5] OWASP lists command injection as MCP5:2025

  • quote: "MCP5:2025 Command Injection & Execution. Command injection in MCP environments occurs when an AI agent constructs and executes system commands"
  • url: https://owasp.org/www-project-mcp-top-10/
  • source_type: primary
  • published: 2025
  • confidence: high

[6] Cloudflare terminates TLS, can inspect all traffic

Dead ends

  • Could not determine if MCP servers have authentication beyond URL secrecy
  • No evidence of secrets scanning in CI pipeline

Suggested follow-ups

  • Check if git history has been cleaned of leaked secrets
  • Audit MCP server authentication mechanisms