The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted source code provided by the user.
Ingestion points: Reads source code files from the user-specified <source-path> referenced in the skill instructions.
Boundary markers: The skill does not implement delimiters or specific instructions to the agent to disregard instructions found within the code comments or content.
Capability inventory: The skill has access to tools such as Write, Bash (via npx), and mcp dispatch, which could be misdirected if the agent follows malicious instructions in the input.
Sanitization: No sanitization or validation of the input source code content is described in the skill.

api-docs