The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's instructions and environment section utilize npx to download and execute the latest versions of @redocly/cli and @stoplight/prism-cli. These are recognized industry-standard tools for linting and mocking API specifications, and their inclusion is directly relevant to the skill's stated purpose of contract verification.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data in the form of OpenAPI specifications during the review-contract and evolve-contract workflows. This represents an attack surface for indirect prompt injection if a malicious specification is provided to the agent.
Ingestion points: OpenAPI specification files (typically YAML or JSON) provided for review or evolution tasks.
Boundary markers: The skill mandates a 'lint' gate using Redocly to verify structural integrity before analysis, which acts as a basic validation step.
Capability inventory: The agent is instructed to run command-line tools (npx @redocly/cli, npx @stoplight/prism-cli) to validate and mock the provided specifications.
Sanitization: The skill relies on the structural validation performed by external linting tools to filter malformed or non-compliant input before the agent processes it.

api-design