The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses perl to perform global text normalization (replacing em-dashes with hyphens) across the project files.
[REMOTE_CODE_EXECUTION]: The workflow involves executing npx eslint, npx prettier, and npm test. These commands execute scripts defined within the local project's package.json, which is a standard procedure for code verification but relies on the integrity of the project being refactored.
[PROMPT_INJECTION]: The skill is configured to read instructions from a local temporary file at /tmp/$(basename "$PWD")/code-review.md. This represents an indirect prompt injection surface where external data could potentially influence the agent's refactoring logic.
Ingestion points: SKILL.md (Workflow Step 2) reads /tmp/$(basename "$PWD")/code-review.md.
Boundary markers: None identified; the skill is instructed to fix findings listed in the file.
Capability inventory: File modification via perl, execution via npx and npm test.
Sanitization: None; the skill treats the file content as actionable review findings.

javascript-refactor