The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses glab (GitLab CLI) and git apply to interact with the Drupal project infrastructure. These operations are scoped to a specific, well-known host (git.drupalcode.org) and follow established developer workflows.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from GitLab job logs and patch artifacts.
Ingestion points: Data is ingested through glab ci trace (job logs) and glab api (patch artifacts) in SKILL.md (Step 3).
Boundary markers: The skill lacks explicit boundary markers for log output, though it enforces strict validation on identifiers.
Capability inventory: The skill possesses network read capabilities (via glab) and filesystem modification capabilities (via git apply).
Sanitization: The instructions include mandatory regex validation (^[0-9]+$ and ^[A-Za-z0-9_./-]+$) for Merge Request numbers and project paths before any API calls are made, which helps mitigate injection via malformed identifiers.

contribution-pipeline