The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and modify project codebases, which creates an inherent attack surface for indirect prompt injection where malicious instructions could be embedded in the files the agent processes.
Ingestion points: The skill uses tools like Read, Grep, and Glob to ingest content from the project filesystem (defined in SKILL.md).
Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the agent when processing external project data.
Capability inventory: The skill has access to Read, Write, Edit, Grep, Glob, Bash, and WebSearch, which allows for significant modifications to the user's environment based on ingested content.
Sanitization: No explicit data sanitization, escaping, or validation logic is provided within the instructions to mitigate instructions embedded in data.

hotwire-coder