recoup-content-write-caption
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes artist context and social media posts, which are external data sources that could be used for indirect prompt injection.\n
- Ingestion points:
context/artist.md,context/audience.md, and social media posts fromapi.recoupable.dev.\n - Boundary markers: The skill instructions include creating a "voice fingerprint" and checking against an "anti-slop checklist."\n
- Capability inventory: Reading local files, writing to the
content/captions/directory, and network access viacurl.\n - Sanitization: Verification of generated captions against specific artist style constraints.\n- [COMMAND_EXECUTION]: The skill uses shell commands like
curl,jq, andsedto interact with the Recoup platform's API for resolving artist IDs and retrieving context data.\n- [DATA_EXFILTRATION]: The skill transmits authentication credentials (RECOUP_API_KEY,RECOUP_ACCESS_TOKEN) to the vendor's official domain (api.recoupable.dev). This is documented behavior for authenticating with the platform.
Audit Metadata