recoup-platform-build-os
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill generates and executes Python scripts (
doctor.py) and shell commands for workspace health checks and data reconciliation. These are core, documented features for the 'self-managing' workspace design. - [PERSISTENCE]: The skill provides templates to schedule the 'janitor' task via GitHub Actions, cron, or launchd. This is the intended primary purpose of the 'never-stale' design, intended to keep workspace state current without manual intervention, and is transparently presented to the user.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data (e.g., meeting transcripts, notes) to seed a knowledge base. While this presents an attack surface for indirect prompt injection, the skill implements safety measures including a 'draft — confirm' tagging system for inferred content and a mandatory approval step for promoting repeatable work into new skills.
- [EXTERNAL_DOWNLOADS]: The skill integrates with external vendor resources including official Recoup API endpoints and associated
recoup-*platform skills as part of its intended music-label management functionality.
Audit Metadata