The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and review user-provided Solidity contracts and dApp source code, creating an attack surface where malicious instructions could be embedded in code comments or data.
Ingestion points: Solidity smart contracts (.sol), dApp frontend code, and Foundry test files provided by the user in the conversation context.
Boundary markers: Not explicitly defined in the skill instructions to separate untrusted user code from the agent's internal logic.
Capability inventory: The skill guides the agent to use developer tools including forge, cast, anvil, and package managers like npx and yarn.
Sanitization: The skill correctly recommends running security audits in a 'separate agent / fresh context' to mitigate bias and potential injection risks.
[CREDENTIALS_UNSAFE]: The instructions contain a strong defensive rule against secret exposure, noting that AI agents are a frequent source of leaked private keys and API keys, and mandating the use of .env files and placeholders.
[EXTERNAL_DOWNLOADS]: The skill references industry-standard tools and protocols, including OpenZeppelin libraries, Scaffold-ETH 2 (create-eth), and specialized agent-era protocols like x402.

ethereum