readme-craft
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data including README files, manifests, and source code entry points, while maintaining the capability to modify project documentation. This creates a potential surface for indirect prompt injection attacks where malicious instructions hidden in a project's files could attempt to influence the agent's behavior.
- Ingestion points: The skill reads existing README files, project manifests (e.g., package.json, Cargo.toml), licenses, entry points, and CI configuration files (SKILL.md).
- Boundary markers: There are no specific instructions or delimiters mentioned to isolate untrusted input from the agent's primary instructions.
- Capability inventory: The skill can perform audits and apply modifications to the project's documentation using the
project-knowledgeapply-method (SKILL.md). - Sanitization: No explicit sanitization, filtering, or validation logic is described for the text ingested from the project files.
Audit Metadata