The Agent Skills Directory

[SAFE]: The skill implements standard DeFi integration patterns for Aave V3. No malicious behavior, prompt injections, or unauthorized data access were identified during analysis.
[SAFE]: Data exposure analysis shows the skill utilizes an environment-provided WALLET_SERVICE_URL and interacts with well-known public RPC nodes (e.g., LlamaRPC, Arbitrum, Polygon) for blockchain queries. This is standard practice for decentralized applications.
[SAFE]: Prompt injection analysis confirmed that instructional language in the documentation is used solely for guiding user interaction and setting up necessary wallet policies, with no attempts to bypass safety filters or override system instructions.
[SAFE]: Dependency and execution analysis found only standard library imports and internal utility references. There is no evidence of remote code execution, dynamic package installation, or unsafe deserialization.
[SAFE]: The skill uses a whitelist-based approach for resolving chain IDs and token addresses, which mitigates risks associated with indirect prompt injection or providing the agent with malicious contract addresses.

aave