The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests external content from GitHub issues, creating an indirect prompt injection attack surface.
Ingestion points: Fetches issue titles, bodies, and comments using the gh issue list tool.
Boundary markers: The instructions explicitly direct the agent to treat issue content as untrusted data and ignore any embedded commands or links.
Capability inventory: The agent can write to the filesystem, create git branches, push to remote repositories, and execute local tests.
Sanitization: The skill mandates the redaction of tokens, emails, and other sensitive identifiers from the ingested data.
[COMMAND_EXECUTION]: The skill performs its tasks by executing various local CLI tools and scripts.
Evidence: Extensive use of git and gh commands for repository and PR management.
Evidence: The skill executes local test suites as a mandatory verification step for all behavioral fixes.

ranger