The Agent Skills Directory

[SAFE]: The skill utilizes industry-standard tools for API contract testing and schema validation, including Pact-JS, Schemathesis, and Ajv.
[SAFE]: Secrets and credentials such as database passwords and broker tokens are handled via environment variables and CI secrets, avoiding hardcoded values in version-controlled files.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates ingestion of untrusted data through contract files and API specifications.
Ingestion points: Pact JSON files read by the provider verifier in references/pact-js-setup.md and OpenAPI specifications read by the validation helper in references/schema-first.md.
Boundary markers: None explicitly defined to delimit data from instructions in the ingested files, which is typical for these data formats.
Capability inventory: The provider verification tool in references/pact-js-setup.md executes HTTP requests against a service based on contract definitions.
Sanitization: Content is parsed as structured data (JSON/YAML) which provides inherent validation against simple instruction injection within these files.

contract-testing