create-context-skill
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and process raw, verbatim error strings, build logs, and external documentation (Phase 1
- Harvest). This creates a potential surface for indirect prompt injection if build logs contain adversarial content designed to influence the agent's summarization or structure.
- Ingestion points: Build logs, verbatim converge failures, and review findings specified in Phase 1 (Harvest).
- Boundary markers: Not specified in the instructions.
- Capability inventory: Execution of validation commands via
npx(Phase 4). - Sanitization: No explicit sanitization or filtering of error logs is mentioned.
- [COMMAND_EXECUTION]: The skill utilizes
npx skills-ref validateto verify the generated skill directory. This is a standard command for the vendor's ecosystem used to ensure compliance with the specified standard. - [EXTERNAL_DOWNLOADS]: Mentions the vendor's official GitHub repository at
github.com/getcolors/workspaceas the source of the standard's definition. This is a reference to a well-known vendor resource.
Audit Metadata