sonamu-entity

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a series of shell commands using the pnpm sonamu CLI to synchronize schemas, generate database migrations, and scaffold application code. These commands are essential for the framework's operational workflow.
  • [INDIRECT_PROMPT_INJECTION]: The skill operates on user-controlled entity.json files that define database schemas and application logic. A malicious configuration could potentially exploit the agent's ability to generate SQL or TypeScript code, although the framework includes built-in validation. 1. Ingestion points: Reads and modifies src/application//*.entity.json and src/application//*.types.ts files. 2. Boundary markers: The EntityManager validates JSON files against a strict schema, rejecting unrecognized keys and invalid types. 3. Capability inventory: The skill can write files, execute CLI commands (pnpm sonamu), and trigger project builds (pnpm build). 4. Sanitization: Employs Zod for schema validation and strictly defined prop types to limit input scope, though it allows raw SQL in specific fields like dbDefault and index predicates.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 11:26 AM