copilot-review-loop
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill identifies and executes project-specific safeguards (compilation, tests, linting) found in files such as
Makefileor.github/workflows/. This involves the agent executing commands defined within the repository files to verify code changes.- [INDIRECT_PROMPT_INJECTION]: The skill processes pull request comments fetched via the GitHub API to determine necessary code modifications. While it filters forcopilot[bot], it processes external data to drive its automated fix-loop.- [INDIRECT_PROMPT_INJECTION]: Mandatory Evidence Chain for project safeguard execution: 1. Ingestion points: Local repository filesMakefile,README.md, and.github/workflows/(Step 1). 2. Boundary markers: Absent. 3. Capability inventory:Bash,git, and execution of shell commands defined in project files. 4. Sanitization: None.
Audit Metadata