comment-to-offer-automation

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The repository contains shell scripts (tests/validate_repo.sh, tests/test_negative_fixture.sh) designed for repository validation and testing. These scripts utilize standard system utilities including grep, sed, ripgrep (rg), and python3. The scripts are intended for developer use to ensure the skill follows formatting and security best practices (e.g., scanning for accidentally included API keys).
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from Instagram (post themes and user comments) to generate automation specifications.
  • Ingestion points: Instagram Reel/Post topics and user comments provided by the user in the prompt.
  • Boundary markers: The skill explicitly instructs the agent to generate specifications ("規格") and documentation only, with clear warnings that it should not log in, enable, or modify any external systems without authorization.
  • Capability inventory: The skill possesses shell execution capabilities through its test scripts (validate_repo.sh) and uses python3 for JSON validation.
  • Sanitization: The validate_repo.sh script actively scans the repository for credential patterns (AWS, GitHub, OpenAI) and forbids the generation of fake URLs or deceptive marketing claims.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 06:57 PM
Security Audit — agent-trust-hub — comment-to-offer-automation