web3-security-tooling

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection vulnerability surface. The skill instructions require the agent to fetch a resource list from an external, untrusted source which may contain hidden instructions.
  • Ingestion points: Remote file https://raw.githubusercontent.com/gmh5225/awesome-web3-security/refs/heads/main/README.md referenced in SKILL.md.
  • Boundary markers: None specified; the agent is not instructed to ignore embedded instructions in the external content.
  • Capability inventory: The agent is tasked with organizing and updating a project's README.md, which involves write operations to the local filesystem.
  • Sanitization: No sanitization or validation logic is defined for the fetched content.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill identifies an external GitHub repository (gmh5225/awesome-web3-security) as a mandatory data source. This repository is not on the trusted sources list, posing a risk that the ingested data could be tampered with to exploit the agent's logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 07:20 AM