The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently uses shell commands to interact with the user's environment, including commands like node -e to parse project data and npm run to execute project-defined scripts.
[COMMAND_EXECUTION]: The diagnostic script scripts/ts_diagnostic.py utilizes subprocess.run(shell=True) to execute system commands, which is a sensitive pattern for potential command injection if project data is malicious.
[REMOTE_CODE_EXECUTION]: The instructions direct the agent to use npx to execute various utilities such as ts-migrate, typesync, vitest, and tsx. This involves downloading and executing packages from the npm registry at runtime.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted project files and executes commands based on their content.
Ingestion points: Project configuration files (package.json, tsconfig.json) and source code in the src/ directory.
Boundary markers: None present; the agent treats project data as trusted instructions for build and test steps.
Capability inventory: Shell execution (npm run, npx), file reading, and subprocess management in scripts/ts_diagnostic.py.
Sanitization: There is no evidence of sanitization or validation of the scripts defined in package.json before they are executed via npm run.

typescript-expert