The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill authors integration contracts (OpenAPI, Pydantic, TypeScript) by processing external input from project plans and configuration files. A malicious plan could influence the resulting code-level specifications to introduce logic flaws or architectural vulnerabilities in downstream implementation agents. \n- Ingestion points: The skill reads docs/agents/contract-format.md and the build plan content. \n- Boundary markers: There are no explicit delimiters or instructions to prevent the agent from obeying malicious instructions embedded within the processed plan. \n- Capability inventory: The skill possesses filesystem capabilities (Read, Write, Edit, Glob, Grep) and executes the tsc --noEmit command to validate TypeScript contracts. \n- Sanitization: No sanitization, escaping, or schema validation is performed on the plan data before it is interpolated into the contract templates.

contract-author