The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the glab CLI and git to interact with GitLab and the local repository. Specifically, the ci.md workflow implements a background monitoring agent that can automatically modify code and perform a git push --force-with-lease to address CI failures. While this is a documented feature, it grants the agent significant autonomy over the codebase.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources that are not fully under the user's control, creating a surface for indirect prompt injection attacks.
Ingestion points: pull-comments.md fetches user-provided discussion threads from GitLab merge requests; ci.md retrieves CI job traces (logs) using glab ci trace.
Boundary markers: The workflows do not explicitly define delimiters or 'ignore' instructions when presenting this external data to the agent, which may lead the agent to follow instructions embedded within comments or logs.
Capability inventory: The skill possesses high-privilege capabilities including writing files, making GitLab API requests (POST/PUT), and pushing code changes to the remote repository.
Sanitization: There is no evidence of sanitization or filtering of the fetched content before it is processed by the agent.

glab