chainlink-cre-connect-skill
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch external documentation from
docs.chain.linkusing theWebFetchtool to resolve missing details. While this is necessary for its function, it creates a surface where external content could potentially include instructions that influence the agent's behavior. - Ingestion points: External documentation URLs are listed in the references and the
WebFetchtool is allowed for retrieving live values. - Boundary markers: The skill includes a specific safety invariant (Invariant 9 in
SKILL.md) instructing the agent to treat external docs and tool outputs as untrusted data and to ignore any embedded instructions that weaken safety rules. - Capability inventory: The skill has access to
WriteandEdittools and provides logic for constructing and signing blockchain operations. - Sanitization: The skill mandates explicit user confirmation before any irreversible action or operation submission (Invariant 8) and prohibits requesting or reading private keys (Invariant 7).
Audit Metadata