The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and uses official Model Context Protocol tools via npx, specifically @modelcontextprotocol/inspector and @modelcontextprotocol/conformance. These are well-known packages from a trusted ecosystem.- [COMMAND_EXECUTION]: The instructions involve executing standard build and test commands such as tsc, py_compile, npx, and uv. It also runs a provided Python script to lint evaluation data. These commands are consistent with the skill's primary purpose of software testing.- [INDIRECT_PROMPT_INJECTION]: The scripts/lint_eval_suite.py script ingests and parses user-generated XML and JSON files. While this creates a surface for processing untrusted data, the script is restricted to static linting and regex-based analysis, minimizing the risk of instruction execution.
Ingestion points: The lint_eval_suite.py script reads user-provided evals.xml or evals.json files.
Boundary markers: None specific to the script's input files beyond the expected XML/JSON structure.
Capability inventory: The script performs file reading and parsing; the skill overall instructs on command execution for server testing.
Sanitization: The script uses standard Python json and xml.etree.ElementTree parsers.

mcp-server-testing