The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands specifically for repository management. It employs git rev-parse, git fetch, git pull, and git stash to ensure the local codebase is synchronized with the remote origin before performing an audit.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process arbitrary source code from a repository.
Ingestion points: Source files within the targeted repository directory or current diff (SKILL.md, Phase 0 and 1).
Boundary markers: The instructions do not define specific delimiters to wrap analyzed content or explicit 'ignore' directives for embedded instructions within the code being audited.
Capability inventory: The skill can execute shell commands (git) and perform file system writes (CLEAN_CODE_AUDIT.md and CLEAN_CODE_AUDIT.html).
Sanitization: The skill focuses on generating report findings and provides guidance in references/html-report-guide.md for escaping HTML characters when populating the visual report template.
[SAFE]: The skill explicitly forbids modification of the source code being audited, ensuring it acts solely as a read-only analysis tool.
[SAFE]: No external network dependencies or remote scripts are downloaded during execution; all reference materials and HTML templates are provided as local files within the skill package.
[SAFE]: The metadata and documentation are consistent with the tool's primary purpose, with no evidence of obfuscation or hidden logic.

clean-code