The Agent Skills Directory

[COMMAND_EXECUTION]: The bundled script scripts/inspect_pr_checks.py executes standard git and gh (GitHub CLI) commands using Python's subprocess.run to retrieve repository metadata and CI status. These calls use argument lists which prevent shell injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from GitHub Actions logs, creating a potential surface for indirect prompt injection attacks.
Ingestion points: The fetch_check_log and fetch_run_log functions in scripts/inspect_pr_checks.py retrieve raw log data from GitHub's servers.
Boundary markers: The extracted log snippets are printed directly to standard output and presented to the agent without specific delimiters or instructions to ignore embedded commands within the logs.
Capability inventory: According to the Workflow section in SKILL.md, the agent is authorized to perform code changes and execute tests based on the analysis of these logs.
Sanitization: The script performs no filtering or sanitization of the log content before it is incorporated into the agent's context.

gh-fix-ci