claude-md
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes user-provided
$ARGUMENTSto determine its behavior (create, update, or audit commands). While standard, this allows untrusted data to enter the agent's decision-making process. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze existing
CLAUDE.md,README.md, and other documentation files. If these files contain malicious instructions, the agent might follow them during the 'update' or 'audit' phases. - Ingestion points: Reads
./CLAUDE.md,.claude/CLAUDE.md,**/CLAUDE.md,~/.claude/CLAUDE.md,README.md,CONTRIBUTING.md,package.json,pyproject.toml,Cargo.toml, etc. (SKILL.md, Sections 1 & 7). - Boundary markers: None explicitly defined in the instructions to distinguish between the content of the files and the agent's instructions.
- Capability inventory: Performs file system reads and writes to create or update
CLAUDE.mdandAGENTS.md(SKILL.md, Output Format section). - Sanitization: No specific sanitization or validation of the input files is mentioned beyond a length recommendation.
Audit Metadata