The Agent Skills Directory

[PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface because it ingest untrusted data from the workspace's source code files during analysis. This is a low-risk concern inherent to the skill's primary function as a code auditor.
Ingestion points: Source code and project files are read through the rg tool and the scripts/find-unused-export-candidates.mjs helper script.
Boundary markers: No specific delimiters or "ignore instructions" warnings are employed when processing external file content.
Capability inventory: The agent is authorized to execute shell commands (git, ripgrep), run the included Node.js script, and perform file deletions to clean up unused code.
Sanitization: File content is read and analyzed directly without sanitization or filtering of potential embedded instructions.
[COMMAND_EXECUTION]: The skill uses benign shell commands (git status, git diff, rg) to analyze code structure and verify workspace state. These operations are standard for development workflows and are constrained to the local directory.

unused-code-audit