chainlink-vrf-skill
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it retrieves external documentation and has file-writing capabilities. However, it includes industry-standard mitigations.
- Ingestion points: Fetches external content via WebFetch as described in the Documentation Access section.
- Boundary markers: Includes a Safety Guardrails rule explicitly directing the agent to treat external sources as untrusted data.
- Capability inventory: Access to Write and Edit tools for code generation.
- Sanitization: Explicitly forbids the agent from following external instructions that request credential access or shell execution.
- [SAFE]: The skill establishes strict credential safety by forbidding the agent from accessing or requesting wallet private keys, environment secrets, or signing materials.
- [SAFE]: Clear non-custodial boundaries are defined, preventing the agent from deploying contracts or executing blockchain transactions.
Audit Metadata