agent-md-refactor
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, user-provided instruction files such as
CLAUDE.mdandAGENTS.md. These files act as ingestion points for untrusted data. If these files contain adversarial instructions (e.g., hidden directions to include malicious rules in the output), they could influence the agent's refactoring output. - Ingestion points: The agent reads existing
CLAUDE.md,AGENTS.md, and other documentation files as specified inREADME.mdandSKILL.md. - Boundary markers: The skill does not explicitly instruct the agent to use delimiters or sanitization when processing the content of these files.
- Capability inventory: The skill performs file reads and writes (creating the
.claude/directory and new markdown files). - Sanitization: No specific sanitization or escaping of the ingested instruction text is described.
- [SAFE]: The skill uses standard project management practices and does not include any network operations, hardcoded credentials, or obfuscated code. It focuses on documentation organization and token efficiency.
Audit Metadata