revise-claude-md
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the transfer of session-specific context into AGENTS.md or .claude.local.md files, which are utilized as context for future sessions. This allows potentially untrusted session content or malicious instructions to be promoted to permanent configuration.
- Ingestion points: The skill analyzes current session history and identified learnings as described in SKILL.md.
- Boundary markers: The skill does not implement delimiters or warnings to isolate the new content from the agent's instructions in future sessions.
- Capability inventory: The skill uses the 'find' command to search the filesystem and performs file write operations to apply changes.
- Sanitization: The potential for injection is mitigated by a human-in-the-loop step requiring explicit user review and approval before any changes are written to disk (Step 5).
Audit Metadata