project-claude-config
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository data such as READMEs, manifests, and file structures to generate configuration audit findings and scaffolding stubs. \n
- Ingestion points:
SKILL.md(Step 2 - Detect pass using Explore subagent to report repo facts and content excerpts). \n
- Boundary markers: The skill includes a "Propose before writing" logic and requires explicit user approval before changes are applied (Step 5). \n
- Capability inventory: The skill can write to instruction files (
CLAUDE.md,AGENTS.md) and modify the settings harness via theupdate-configskill. \n - Sanitization: All proposed changes are presented as a ranked synthesis for manual user review and approval before execution. \n- [EXTERNAL_DOWNLOADS]: Fetches official configuration documentation from
code.claude.comto validate that proposed harness settings, hooks, and permissions are current and non-deprecated. This targets a well-known service associated with the agent's platform and is part of the skill's primary auditing function.
Audit Metadata