The Agent Skills Directory

Category 8: Indirect Prompt Injection (LOW): The skill is designed to ingest and process data from external APIs, which creates a surface for indirect prompt injection. However, the skill provides extensive documentation on boundary markers and sanitization, specifically recommending the use of Zod for schema validation and strict type transformations at the integration boundary (e.g., in client-request-response-caching.md).
Ingestion points: ApiClient.request and UserService methods in client-base-service-layer.md and client-request-response-caching.md.
Boundary markers: Explicitly recommends Zod schemas for parsing and validating untrusted data.
Capability inventory: Network operations via fetch for API communication.
Sanitization: Usage of zod for input validation and manual transformer functions to convert API shapes to internal domain models.
Category 2: Data Exposure & Exfiltration (SAFE): The skill demonstrates safe handling of credentials. It recommends loading keys from environment variables (process.env.API_KEY) and provides a redactHeaders helper in client-request-response-caching.md to prevent logging sensitive information like authorization tokens or API keys.
Category 4: Unverifiable Dependencies (SAFE): The skill references well-known, industry-standard tools and libraries for API development, such as zod, msw, nock, openapi-generator, and orval. No suspicious remote script executions or untrusted package sources were identified.

api-integration