The Agent Skills Directory

[SAFE]: The skill provides legitimate development instructions for the Apify MCP server project and its associated evaluation framework.
[EXTERNAL_DOWNLOADS]: The instructions refer to the use of official vendor tools and services, including the apify-client and langfuse-cli packages, as well as the langfuse.apify.dev platform. These are well-known services associated with the skill's author (apify).
[CREDENTIALS_UNSAFE]: The documentation references the use of environment variables for API keys such as APIFY_TOKEN and LANGFUSE_SECRET_KEY. It follows standard security practices by instructing users to store these in a local .env file and utilizes non-sensitive placeholders in all examples.
[COMMAND_EXECUTION]: The skill involves executing local development commands and scripts (pnpm run, tsx, npx) which are typical for software engineering evaluation and testing workflows.
[INDIRECT_PROMPT_INJECTION]: The skill describes a test harness where an agent processes user-language queries and interacts with external web content. This is the intended purpose of the evaluation system, and the documentation provides guidance on handling expected errors and structured validation to mitigate reliability issues.

creating-mcp-agent-evals