chainlink-ace-skill

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs the agent to fetch and use live content from the public repo (https://github.com/smartcontractkit/chainlink-ace and its raw base https://raw.githubusercontent.com/smartcontractkit/chainlink-ace/main/) and the product docs (https://docs.chain.link/ace) at runtime as the authoritative source of truth, meaning those URLs are fetched during runtime and their content directly controls the agent's prompts/answers.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill is explicitly about on-chain contract operations and token/compliance management (self-deploying ACE contracts, ERC-20 and ERC-3643 token examples, issuing/revoking credentials, registering targets/identities, deploying/configuring/upgrading PolicyEngines). The SKILL includes an Approval Protocol that prepares and then (with second confirmation) executes write actions that deploy/configure/register/issue/revoke or otherwise write onchain state. Those are specific blockchain transaction capabilities (crypto/blockchain) capable of affecting funds/tokens. Therefore it meets the "Crypto/Blockchain (Wallets, Swaps, Signing)" category of direct financial execution.