issue-autopilot
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI ('gh') to perform branch creation, code commits, and pull request generation as part of its completion workflow.
- [COMMAND_EXECUTION]: Subagents are instructed to execute project-specific build, lint, and test commands (e.g., 're-run the relevant tests/build/lint') to verify the quality of implemented changes.
- [DATA_EXFILTRATION]: The skill transmits local code and task status information to GitHub repositories when creating branches and pull requests as intended for its autonomous workflow.
- [PROMPT_INJECTION]: The skill processes data from local markdown issue files, creating an indirect prompt injection surface. (1) Ingestion points: Local issue files in the target folder identified in Phase 0. (2) Boundary markers: Absent; issue paths are passed to subagent prompts without explicit delimiters or warnings to ignore embedded instructions. (3) Capability inventory: The skill and its subagents have the ability to modify the local file system, execute shell commands via 'execute-issue', and perform Git operations. (4) Sanitization: No sanitization of issue file content is performed before it is passed to autonomous subagents.
Audit Metadata