bootstrap-api-keys
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by auditing existing project files (e.g., migrations, schemas) and accepting user-supplied configuration values to customize the generated code.
- Ingestion points: Discovery phase checks using
GlobandGrepon application directories, and user configuration prompts for parent associations and route paths. - Boundary markers: No explicit boundary markers are defined to isolate external data from the agent's instruction context.
- Capability inventory: The skill writes new Elixir source files and database migrations to the filesystem.
- Sanitization: The skill uses template interpolation for code generation; while standard for this use case, it does not describe specific sanitization for the inputs retrieved from the existing project context.
- [SAFE]: The skill incorporates strong security defaults for the generated infrastructure. It ensures that raw API tokens are never stored in the database, requiring the use of SHA256 hashes instead. It also implements a 'show once' policy for tokens and uses a prefix convention to mitigate lookup timing attacks and accidental leakage.
Audit Metadata