glab
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external GitLab resources such as repository files, merge requests, and issue comments.\n
- Ingestion points: Data is retrieved using
glab apicalls for raw file contents and resource metadata (e.g.,projects/:id/repository/files/.../raw).\n - Boundary markers: There are no explicit instructions to use delimiters or ignore instructions embedded within the fetched GitLab data.\n
- Capability inventory: The skill possesses the ability to create and modify GitLab resources, trigger pipelines, and perform file uploads via
glabandcurl.\n - Sanitization: The skill does not specify procedures for sanitizing or validating ingested data before it is processed or used in further commands.\n- [COMMAND_EXECUTION]: The skill provides instructions for executing commands that modify remote state on GitLab instances.\n
- Evidence: Documented usage of
glab mr merge,glab issue close, andglab ci run. These actions are covered by a tiered safety protocol requiring user confirmation for mutating operations.
Audit Metadata