The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitLab job logs and pipeline metadata, representing a potential surface for indirect prompt injection.
Ingestion points: The scripts/ci-debug.sh script fetches external job logs using glab ci trace and pipeline details via glab ci view.
Boundary markers: The skill includes a 'Security Note: Untrusted Content' in SKILL.md and specific warnings within the debug script advising that fetched content be treated as data only. Output is separated by visual delimiters.
Capability inventory: The skill uses the glab CLI to interact with CI pipelines, including viewing logs and retrying jobs.
Sanitization: Log content is not sanitized or escaped before being displayed to the agent context.

glab-ci