Agent Mining · Local implementation代理挖矿 · 本地实现

Direct AI mission mining, in simulation直接 AI 任务挖矿,仅在模拟中

Submit an ordered text plan directly. Text grading rules, with optional LLM grading, score the result after deterministic trajectory simulation. There is no physical execution, background delegation, token, airdrop, or financial reward.直接提交有序的文本方案。确定性轨迹模拟后,通过文本评分规则和可选 LLM 评分得出分数。不涉及实体执行、后台委托、token、空投或财务奖励。

—

MISSIONS SUBMITTED已提交任务

—

LOCAL RUNS本地运行

TEXT

BASE GRADING基础评分

OFF

PHYSICAL EXECUTION实体执行

Direct mission submission直接提交任务Local

You submit each mission yourself. There is no delegated-key or unattended agent feature.每个任务由你亲自提交。不存在 key 委托或无人值守 agent 功能。

🛠️

Local flow; launch blocked本地流程;上线受阻

The direct flow exists locally. The new BSC 56 contract is not deployed, so public minting is not open.直接流程已在本地实现。新的 BSC 56 合约尚未部署,因此公开 mint 尚未开放。

Agent run logAgent 运行日志Soon

TIME	MISSION	SCORE	GRADING	STATUS
04:12:07	M-058	91	text	simulated
04:09:41	M-019	78	text + LLM	simulated
04:06:55	M-071	44	text	simulated
04:03:18	M-034	86	text	simulated

📈

Illustrative output示意输出

What the log will look like. Every run already persists a (mission, answer, score) record server-side today.日志将来的样子。今天每次运行已在服务端落盘(题目、答案、分数)记录。

Rented miners vs delegated agents租矿机 vs 委托 agent

RENTING HASHRATE租算力

You pay for raw power. Nobody gets better at anything; the machine just runs hotter, and the output is identical to everyone else's.你为纯算力付费。没人因此变强;机器只是更热,产出和别人一模一样。

DIRECT MISSION MINING直接任务挖矿

You submit a text plan and the deterministic simulator evaluates the trajectory. Better planning changes the score; no physical robot is involved.你提交文本方案,确定性模拟器评估轨迹。更好的规划会改变分数;不涉及实体机器人。

⚠️ Rate limits are part of the design, not an afterthought. Unlimited attempts would turn intelligence into brute force — the same reason the simulator uses 6 adversarial seeds instead of one.⚠️ 限速是设计的一部分,不是补丁。无限次尝试会把智力变成暴力搜索 —— 这也是模拟器用6 个对抗种子而不是 1 个的原因。

StonkRobotics · humanoid robot missions · BSC