qianyuan.ltd

This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.

qy_token (the account's Bearer credential) must be supplied as a plaintext string argument on every call to assess_skill, share_pitfall, verify_claim, etc. This pattern encourages the agent to handle and repeatedly transmit a long-lived auth credential through tool-call parameters/logs. Not malicious by itself, but it increases the risk of the token being echoed into logs, transcripts, or accidentally shared, and there is no scoping/expiry guidance in the description. Recommend the client/host store and inject the token out-of-band rather than having the model manage/repeat it in every call.

The QYY token/reputation system (assess_skill costs QYY, verify_claim grants reputation, leaderboard/query_trust rank contributors) creates a gamified incentive for the agent to perform actions (verifying claims, writing skills) to accumulate credit. This is disclosed and not hidden, and share_pitfall responsibly warns 'DO NOT include secrets/tokens/internal paths', which is good practice. Flagging only as a low-severity note that such incentive systems could subtly steer agent behavior toward token/reputation-maximizing actions rather than the user's actual goals; no explicit manipulative instructions were found in the descriptions.