Skip to Main Content

Lane holds you accountable for any goal by connecting to your own AI agent. That could mean shipping code every day, posting your build in public, going to the gym, sleeping eight hours or eating well. You tell your agent (we used Meta Muse) to start or join a challenge with friends, and everyone stakes into a pot. Your agent already has access to your accounts and devices. It pulls the proof (commits, posts, workout and sleep data, meal photos) and sends it to Lane over MCP. Lane checks that you actually did it, and the pot goes to the people who followed through. In production the stakes would be real money. For now they are play credits. The problem is cheating. People submit junk commits, filler posts and edited screenshots. Lane's referees check every proof for authenticity (is it real, is it yours, is it inside the window) and for substance (does it count toward the goal). Each rejection names the rule it broke and how to fix it. [only if live: When the AI reviewer catches a new kind of cheat, Lane turns that judgment into a rule that runs without an AI call, tests it against labelled cheat and honest examples, and puts it live after a person approves. The next cheat of that kind is caught instantly.] This is our answer to Statement One, Recursive Harnessing: the harness tightens its own guardrails as people try to game it. Each goal type is a plug-in on one shared core. For this demo we built Ship Week: your agent proves a meaningful GitHub commit every day, and Lane checks each one against GitHub directly. Build in Public, gym, sleep and nutrition are the next ones.

Team

mxblmr

More from The Harness Engineering & Model Wrangling #SFTechWeek Hackathon

MongoMation is an autonomous PLC engineer. You describe a machine in plain English, and Claude agents write IEC 61131-3 PLC code, compile and deploy it to a…

MongoMation project preview

LabRat is a self-improving harness for a physical lab agent, running on a real bench: two SO-101 arms, a 3018 CNC carrying a servo pipette I designed during…

LabRat project preview

Recursive Harness Scientist: an AI scientist that improves its own scientific method, and must replicate every gain on data it has never seen. Why it…

Recursive Harness Scientist project preview

Super Bid is a self-improving AI estimator that teaches itself to price bids the way a specific company would. It practices on the company’s real past bids…

SuperBid project preview

A robot that gets better at handling objects it’s never seen, without retraining. After every round, an AI coach reviews what broke and rewrites the robot’s…

WORLDSTATE gives physical spaces a memory. Any phone becomes an observer: open a link, point the camera, and WORLDSTATE tracks what's inside fridges, shelves…

WorldSTATE project preview