trustless-work-arc-escrow

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The workflow reads blockchain logs, escrow metadata, and external work evidence (URLs, pull requests, IPFS files) submitted by third-party workers or disputing parties, representing low-risk exposure where the agent actively fetches selected outsider-authored content.

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

• Direct money access detected (medium risk: 0.30). The skill documents direct financial operations on Arc testnet, including locking USDC, funding escrows, releasing payments, splitting funds in disputes, and executing transactions via Circle CLI (SKILL.md:3, 105-126). However, authentication and authorization are handled externally via the Circle Agent Wallet and Circle CLI (which requires human login and API key management as specified in SKILL.md:35-36, 55, 69-71), and every action is validated and built through the independent Trustless Work API before signing/execution (SKILL.md:13, 40-41, 119-125). Thus, qualifying independent access control is present for these direct financial operations.