The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install additional skill modules from an external source (mattpocock/skills) using the npx skills utility. The instructions include the npm_config_ignore_scripts=true flag, which is a security best practice to prevent the execution of potentially malicious lifecycle scripts.
[COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) and git to retrieve repository metadata, such as determining the default branch and identifying merge bases. These operations are standard for development-focused skills and are used to automate the code review process.
[PROMPT_INJECTION]: The skill is designed to process untrusted data which creates a surface for indirect prompt injection.
Ingestion points: The agent reads git diffs, commit references, and originating issue descriptions during the 'Review until green' phase.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the diffs or issues.
Capability inventory: The skill can modify the codebase via the implement child skill and interact with remote repositories using the gh tool.
Sanitization: There is no evidence of escaping or validation performed on the external content before it is processed by the agent.

polish-branch