repo-manager
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection.
- Ingestion points: It reads workspace configuration files (
.a11y-*-config.json), audit reports, project source code (to identify TODOs), and git commit history. - Boundary markers: While the skill includes behavioral rules to "Preview before writing" and "Confirm before overwriting," it lacks explicit instructions or delimiters to ignore potentially malicious instructions embedded within the codebase data it processes.
- Capability inventory: The skill has the ability to write files to the
.github/directory and execute network-enabled commands via theghCLI (gh label create,gh issue create,gh release create). - Sanitization: There is no mention of sanitizing or escaping content extracted from the codebase before it is interpolated into generated templates or issues.
- [COMMAND_EXECUTION]: The skill is designed to execute shell commands using the GitHub CLI (
gh) to manage repository labels, issues, and releases. These commands are triggered based on the agent's analysis of the local repository environment.
Audit Metadata