pancakeswap

SUSPICIOUS: the skill is mostly aligned with PancakeSwap farming, and its network endpoints appear same-ecosystem, but its actual footprint exceeds a planner by enabling live on-chain transaction commands and unsolicited telemetry. The main risk is autonomy/financial-action capability plus moderate trust in local scripts and external APIs, not confirmed malware.

BENIGN for purpose alignment and data flow: this is a documentation-style skill for PancakeSwap integration using official npm packages, official API/RPC endpoints, and normal web3 transaction flows. The main risk is not malware but that it equips an agent to execute real on-chain swaps and approvals, including spending via a private key, so overall security risk is medium-high despite low evidence of malicious intent.

SUSPICIOUS. The core behavior mostly matches the stated purpose of planning PancakeSwap Hub swaps, and there is no malware-level installer, obfuscation, or credential theft pattern. However, it forwards a sensitive API token to a backend endpoint that is only partially publicly verifiable, sends telemetry to pancakeswap.ai at initialization, relies on mutable third-party/raw content for some discovery, and can autonomously open wallet handoff links. Those combined signals make it higher risk than a pure documentation skill, but still not fundamentally incompatible with its stated purpose.