The Agent Skills Directory

[COMMAND_EXECUTION]: Executes shell commands via gh (GitHub CLI) and glab (GitLab CI CLI) to retrieve pipeline status, logs, and pull request metadata.
[CREDENTIALS_UNSAFE]: Provides recipes to list CI/CD secrets and variables which may contain sensitive environment data.
Evidence: SKILL.md and references/ci-queries.md include gh variable list, gh secret list, and glab variable list.
Evidence: The skill documentation in references/allowlist.md explicitly notes that glab variable list exposes CI/CD variable values such as API keys and passwords.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing external data from CI/CD logs and pull request metadata.
Ingestion points: Command outputs from gh run view --log, gh pr view, and glab ci trace (files: SKILL.md, references/ci-queries.md).
Boundary markers: Absent; command outputs are processed without delimiters or instructions to ignore embedded content.
Capability inventory: Executes shell commands (gh, glab, jq) which could be influenced by malicious content in logs or PR fields.
Sanitization: Absent; output is directly presented to the agent or piped to other utilities.

git-ci