The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a bundled Python script (scripts/inspect_pr_checks.py) to execute gh and git commands for inspecting repository status and CI logs. These operations are performed using subprocess.run with argument lists (avoiding shell=True), which prevents shell injection vulnerabilities.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core function involves reading and summarizing external CI logs that could contain malicious instructions.
Ingestion points: External data enters the agent context via CI logs retrieved from GitHub Actions in scripts/inspect_pr_checks.py (specifically through gh run view --log and the GitHub API).
Boundary markers: There are no explicit markers or instructions used to isolate the log content or tell the agent to ignore instructions found within it.
Capability inventory: The skill has the ability to read files, run specific CLI tools, and suggest/implement code changes to the local repository.
Sanitization: The skill does not perform sanitization or filtering on the retrieved log content.
Mitigation: The workflow includes a critical safety control requiring explicit user approval before any proposed fixes are implemented (Step 7 in SKILL.md), which significantly reduces the risk of autonomous execution of injected commands.

gh-fix-ci