The Agent Skills Directory

External Binary Execution: The skill facilitates the discovery, download, and execution of CI artifacts from GitHub repositories using the gh CLI. This is an intended functionality for validating pull requests, and the skill explicitly recommends security best practices, such as treating PR binaries as untrusted code execution and using isolated test environments.
GitHub CLI Integration: The workflow utilizes the official GitHub CLI (gh) for repository interaction. The use of --jq to filter API responses for specific fields (like headRefOid or id) is a positive pattern that reduces the risk of processing unintended metadata.
Provenance and Integrity Verification: The instructions emphasize verifying source SHAs and merge-commit metadata before execution to ensure the integrity of the binaries being tested.
Indirect Prompt Injection Surface:
Ingestion points: The skill fetches metadata from external repositories using gh pr view and gh api calls in SKILL.md.
Boundary markers: The skill uses structured extraction via --jq to isolate specific identifiers, which naturally limits the processing of untrusted string content.
Capability inventory: The skill possesses the capability to execute shell commands and run downloaded executables, as described in the testing workflow.
Sanitization: The use of structured data filtering through jq provides a level of sanitization by ensuring only expected fields are utilized in the downstream commands.

ci-artifact-testing