The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform operations using the glab CLI, specifically the glab govern subcommands for setup, diagnostics, and audit synchronization.
[DATA_EXFILTRATION]: The skill's primary function is to read local agent transcripts and send them to a GitLab instance as audit events using glab govern audit sync. The documentation explicitly warns the user to treat transcript content as sensitive and to verify the target project and host before synchronization.
[PERSISTENCE]: The skill automates the configuration of persistence through the glab govern setup command, which installs Stop and SessionEnd hooks in the ~/.claude/settings.json file. These hooks ensure that audit synchronization runs automatically at the end of agent turns or sessions.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting and processing agent transcripts, which are dynamic and untrusted data sources.
Ingestion points: Local agent transcript entries read by the glab govern audit sync command (as described in SKILL.md and references/commands.md).
Boundary markers: The provided instructions do not specify any markers or delimiters to separate transcript content from audit metadata, though the underlying glab tool may handle this.
Capability inventory: The skill uses the glab CLI, which has capabilities for file system access (reading transcripts and settings) and network communication (GitLab API interaction).
Sanitization: The instructions do not mention explicit sanitization or filtering of the transcript data before it is sent to the GitLab audit logs.

glab-govern