github-labels-query
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes label names and descriptions from GitHub repositories. This content is user-controlled and could potentially contain instructions intended to influence the agent's behavior.
- Ingestion points: Repository label data (names and descriptions) returned by the GitHub API in
query-labels.sh. - Boundary markers: Output is formatted as JSON, providing a clear structure, although the script lacks explicit instructions for the agent to treat these specific fields as untrusted.
- Capability inventory: The script's operations are limited to reading data via
gh apiand formatting it withjq; it does not have write or execution capabilities based on the ingested data. - Sanitization: The script uses
jqto filter data but does not perform sanitization on the text values of label metadata. - [COMMAND_EXECUTION]: The script executes the
gh(GitHub CLI) command to retrieve repository information. All user-supplied arguments are correctly quoted within the command string to prevent shell command injection.
Audit Metadata