The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external sources, specifically GitHub repository metadata, issue descriptions, and pull request content.
Ingestion points: Data is pulled into the agent context via gh search repos, gh issue list, and gh repo view commands.
Boundary markers: The skill relies on inherited workflow safety from the issue-fixer base skill and advises the agent to read repository-specific instruction files before modification, but does not specify explicit delimiters for external content in this extension.
Capability inventory: The skill utilizes git, gh (GitHub CLI), and local python3 execution for data processing.
Sanitization: Explicit sanitization of issue body text is not detailed in the extension files; the skill focuses on domain-specific logic rather than data parsing safety.
[DYNAMIC_EXECUTION]: The references/repo-discovery.md file includes utility scripts executed via python3 -c to process repository health signals.
Evidence: One discovery script uses eval(line.strip()) to parse JSON-like output generated by the GitHub CLI's jq filter. While eval is generally risky, in this context it processes specific metadata (stars, dates) from a hardcoded list of repositories controlled by the skill author, mitigating the risk of arbitrary code execution from untrusted API responses.

ai-agent-fixer