claude-extensibility
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill handles the creation and optimization of agent and skill definition files (e.g., .claude/agents/, .claude/skills/) based on user-supplied descriptions and goals. This process involves interpolating potentially untrusted user input into persistent system prompt files.
- Ingestion points: User requests to generate, edit, or optimize extensions as described in SKILL.md.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' markers when incorporating user-provided content into generated extension files.
- Capability inventory: The skill specifies actions like 'create, edit, delete extensions', which involves file system write operations.
- Sanitization: There is no requirement for sanitization or validation of user-provided instructions before they are written to the filesystem as persistent system prompts.
Audit Metadata