The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands and a diagnostic Python script (scripts/ts_diagnostic.py) to execute standard development tools such as tsc, npm, and grep for project analysis. The script employs subprocess.run with shell=True for these operations, though the command strings are hardcoded literals.
[INDIRECT_PROMPT_INJECTION]: The skill processes external project configuration and source code, creating a potential surface for indirect prompt injection.
Ingestion points: Processes package.json, tsconfig.json, and source files in the src/ directory.
Boundary markers: None.
Capability inventory: Executes system commands (tsc, npm, grep) and reads local files via standard JSON libraries.
Sanitization: The skill uses static command strings for its shell operations, which prevents the interpolation of untrusted file content into the command execution path.

typescript-expert