workflow-green-repo

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to ingest and act upon data found in repository files which could contain malicious instructions.
  • Ingestion points: The skill reads package.json, Makefile, justfile, CI workflows (.github/workflows/*), AGENTS.md, CLAUDE.md, and README to discover gate commands (Phase 0, Step 2).
  • Boundary markers: No explicit markers or warnings are included to prevent the agent from following instructions embedded within these external files.
  • Capability inventory: The skill utilizes shell execution tools to run the discovered commands during baseline capture (Phase 0) and the final verification phase (Phase 3).
  • Sanitization: The skill does not perform validation or sanitization of the commands extracted from project configuration files before execution.
  • [COMMAND_EXECUTION]: Dynamic Command Discovery and Execution. The skill identifies and executes shell commands dynamically based on the project's environment and configuration files.
  • Evidence: Phase 0, Step 2 and Step 4 direct the agent to identify and run commands for typechecking, linting, and testing found in local files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 11:19 AM
Security Audit — agent-trust-hub — workflow-green-repo