The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes npx, npm, and node to perform project diagnostics, version checks, and validation. Safety is maintained by using the --no-install flag to prevent the execution of remote, untrusted packages. The diagnostic script scripts/ts_diagnostic.py executes commands using argument lists rather than shell strings, reducing the risk of command injection.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from package.json, tsconfig.json, and files within the src/ directory.
Ingestion points: Reads project configuration and source files in SKILL.md (via shell fallbacks) and scripts/ts_diagnostic.py (via grep_source_count).
Boundary markers: The instructions do not define specific delimiters for separating project data from the agent's internal reasoning context.
Capability inventory: The skill can execute shell commands (npx, npm, node) and perform file deletions (rm -rf node_modules/.cache).
Sanitization: The Python script uses subprocess.run with list-based arguments to avoid shell-level manipulation.
[SAFE]: The identified behaviors are standard for a development-focused TypeScript tool. The skill explicitly advises users to protect source excerpts before sharing and uses defensive programming techniques, such as pure-Python file scanning and command timeouts, to ensure safe operation.

typescript-expert