onboarding
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external codebase data, including directory structures, source files, and configuration files, to generate documentation templates. This data ingestion surface could be exploited via indirect prompt injection if the scanned files contain malicious instructions designed to influence the agent's output.
- Ingestion points: Project structure commands (ls, find), entry points (main.go, App.tsx), and configuration files (package.json, pyproject.toml) listed in the SKILL.md file.
- Boundary markers: Output is contained within markdown templates for files like CLAUDE.md and ARCHITECTURE.md.
- Capability inventory: The skill reads files and generates markdown text; it does not invoke network tools or execute scripts based on the ingested content.
- Sanitization: No specific sanitization or filtering of codebase content is defined in the instructions.
- [CREDENTIALS_UNSAFE]: The documentation templates include examples of environment variables such as DATABASE_URL and JWT_SECRET, and instructions for setting up a .env file. These are provided as descriptive placeholders for project configuration and do not represent hardcoded credentials or unsafe storage practices.
Audit Metadata