The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands to perform project diagnostics and validation.
Evidence: The script scripts/ts_diagnostic.py uses subprocess.run(cmd, shell=True) to execute system commands including npx tsc, node -v, and grep.
Evidence: SKILL.md provides instructions for the agent to run npx tsc, npm test, and npx vitest to validate code changes.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic Node.js execution to parse project metadata efficiently.
Evidence: SKILL.md includes a node -e one-liner that executes JavaScript at runtime to extract and join dependency keys from the package.json file.
[INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted project data, creating a surface for indirect prompt injection.
Ingestion points: The agent is instructed to read and analyze package.json, tsconfig.json, and various TypeScript source files (src/**/*.ts).
Boundary markers: There are no explicit boundary markers or instructions to treat data from these files as untrusted or to ignore embedded natural language instructions.
Capability inventory: The skill possesses shell execution capabilities (subprocess.run and npx tools) and file system access, which could be abused if the agent follows instructions found within project files.
Sanitization: No specific sanitization, filtering, or validation is performed on the content read from files before it is processed by the AI or used in subsequent diagnostic operations.

typescript-expert