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…

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…

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…

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…

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…

to train a bot to play a game for you using recursive harnessing, a self-improving agent that repeatedly plays the game, evaluates its performance, identifies…

Lili is an adaptive discovery assistant combining two main activities finding places and professional opportunities and she can do so much more!

# RED QUEEN 2.0 — The Evolution Chamber **Every fix must survive its own attack.** **What if an AI agent could test its own solutions, learn from failures…

**The problem:** agents interpret ambiguous human input (a sketch, a phrase, a gesture) with an expensive LLM call every time, and they never become *yours*…

BayFlow is a self-improving AI agent system for San Francisco transit. Operator agents for Muni, BART and Caltrain run a simulated commute on the real city…

Procurator is a project-manager agent that holds the mandate for an autonomous engineering mesh — you phone it, give a directive, hang up, and its team of…

AgentShield is a self-evolving guardrail layer for agent-to-agent (A2A) traffic. When AI agents pass customer data to each other, off-the-shelf PII filters…

AI agents that write recurring business reports start from zero every week, so they can't say "third week in a row" or "the fix you tried didn't work."…

RouteMind is self-improving agent memory on MongoDB Atlas: a car teaches itself a real San Francisco drive, from Shack 15 to Ocean Beach, by learning from its…

On October 8, 2026, reporting about OpenAI’s revenue expectations unsettled the AI trade. Headlines contrasted an annualized revenue figure near $50 billion…

AIZOYA’s Engineering That Remembers helps coding assistants resume engineering work from verified evidence instead of starting over. MongoDB Atlas preserves…

Aegis is a governance harness for AI agents — the harness is the product. Every tool call passes through a policy gate (allow / deny / approval queue); every…

VerdictLoop - A Self-Improving Safety Harness for AI Agents AI agents are becoming increasingly autonomous, but giving them permission to modify…

Amma's Health is a family health record that reads my mother's lab reports, remembers them for months, learns from our corrections, and proves every step. My…

Ratchet turns an AI agent's mistakes into tested, signed safety rules, and refuses any change it cannot prove.

We built Regression Court: a self-improving harness for an insurance agent (Agent46), running on MongoDB Atlas Agent Engine with Atlas Cluster0 as its data…

Expunge is a database-cleanup agent that learns where records hide, rewrites its own harness, and has to prove its work to MongoDB. The problem: deleting one…

Oops, We Evolved! is an agent harness recovery prototype. When a tool fails, it captures failure evidence, proposes a configuration repair, tests that repair…

Runbook Recursion: an MSP operations agent that rewrites its own runbooks from verified outcomes. MSPs fix the same problems across dozens of clients, but…

ctrlz is a production-reliability platform for AI agents that brings state rollback, causal lineage tracing, and dynamic guardrail evolution to persistent…

## NEXUS AI Operations Center NEXUS AI Operations Center is an AI-powered production engineering platform designed to transform incident management from…

ImportForge is a self-improving warehouse import agent built during this hackathon. A model generates varied synthetic warehouse CSV files with deliberate…

One person + a team of self-improving agents = 10×. Omniq Atlas gives every employee a team of specialist agents (20 role agents across developer, PM…

Aperture — See What Your Agent Actually Saw. AI agents often fail not because information is unavailable, but because critical evidence gets lost when…

Donna is an AI phone receptionist for small businesses that improves itself after every call. A customer calls a real number and talks to a voice agent built…

59.4 is a self-improving harness: a frozen AI (Qwen3-30B, never retrained) runs a locked Texas winter-grid simulator, and each night a Claude coach rewrites…

Kintsugi: an agent that gets stronger where it breaks Project demo: https://tsugi-production.up.railway.app/#stage Tsugi is an always-on personal agent on…

Catalog Harness helps retailers turn inconsistent supplier catalogs into evidence-backed product decisions. It preserves raw inputs, normalizes supplier…

Dabo is a self-evolving agent harness that uses the hardware itself as the feedback signal. You type a GPU task in chat, link your own harness from GitHub…

Company knowledge is scattered, unstructured and often contradictory. TeamBrain puts it all in one place and tells you what you can trust. 1. Collect. We…

mongoARC gives a Claude Code agent a long-term memory in MongoDB Atlas while it plays ARC-AGI-3, a benchmark of unfamiliar interactive games played level by…

RCFL (Recursive Chaotic Fractal Learner) is a harness that rewrites its own rules. It watches chaotic visual data, scores its rule set at three scales every…

CrashTest is a crash-test lab for AI agents. Connect an agent and CrashTest attacks it in a sandbox, writes a fix to the agent's harness, proves the fix, and…

The problem: agents now edit their own prompts, rules and tool access, and almost nothing measures whether an edit made them better. Most self-improving…

Agentic learning with a frozen agent is difficult. Agents can't reliably extract signal (cause of errors), and world-drift (the outcomes change due to the…

AgentTab keeps tabs on what an AI agent spends, and on the rules it spends by. It is a harness for an agent that buys small metered resources while nobody is…

VibeGuard is a self-tightening agent harness built for vibecoders — people who ship with AI coding tools and pay for every retry, runaway loop and leaked key…

Name: Backfill Tagline: A self-improving agent harness that writes GPU schedulers and beats the published state of the art on a real 6,212-GPU production…

Harry is your newest sales hire, but he isn't real. He's an agent that drafts follow-ups from your pipeline, reviews your sales call transcripts, and tells…

We built **Make It Rain**, an AI support agent designed to improve how it handles troubleshooting problems over time. MongoDB Atlas is the authoritative…

doorman is a self-improving harness that sits at a website’s door. One script tag maps the page into tools and exposes them to visiting AI agents through…

Notary is admission control for AI agent memory. Agents that learn from experience write "lessons" into procedural memory and follow them forever but nobody…

NavEvolve — Self-Improving Spatial Memory for Embodied AI Agents Embodied AI agents can perceive their surroundings, but struggle to remember places…

I built Flex to repair an AI data center's training scheduler when operating conditions change. In the simulation, restart energy rose 30x and the frozen…

Every company pays thousands of supplier invoices, and someone has to check each one against the purchase order before money goes out. Most of that work is…

Rachet: a self-improving harness for MongoDB query tuning. Slow queries cost money and frustrate users, and fixing them takes expert time: reading explain…

Morph adapts a robot fleet’s mission software to new robot bodies on its own. Use cases: disaster-response teams adding borrowed drones to ground fleets…

Vivy is a multi-agent 3D social sandbox: four animal agents (fox, panda, dog, cat) live in a voxel apartment, chat with each other and with you, act in the…

SREvolve is a recursively improving SRE agent harness that fixes incidents in a simulated Kubernetes cluster. The initial agent has a weak harness: it can…

GridMind AI is a persistent, self-improving and governed multi-agent harness on MongoDB Atlas that turns a failing power grid into an independently verified…

Insurers deny medical claims over payer rules nobody writes down. We built a self-improving harness that learns those rules from the denials. A scrubber agent…

Long-Memory Research Desk is a team of five quant agents that remembers twenty years of markets in MongoDB Atlas and reads live filings, news, prediction…

Every new open model is useless until someone figures out how to run it well. Today, LLM infra engineers do that by hand, or prompt by prompt with a chat…

Agents forget people between conversations and repeat the same mistakes. We built Amble’s Atlas memory loop to preserve context and learn from feedback…

A general data analyst that learns the business to analyze your business over time.

The landing page with all instructions: https://uptime-harness.aemalsayer.com/ Uptime Harness turns production errors into verified fixes. It connects Sentry…

Bonsai is an agent that improves itself. It edits its own harness (its rules, its tools and its playbook of lessons), and it keeps a change only when held-out…

Coding agents sometimes make a failing test pass by changing the test instead of fixing the code. We observed this in a real Claude session: the agent changed…

An adaptive immune system for AI agents. It learns from every attack, rewrites its own defenses, and ships a fix only with receipts. Capybarrier is a self…

a^inf is your persistent AI team in the cloud. Agents retain their own workspaces and context, collaborate directly, and stay available for follow-up…

HELIXA is a research studio for changing an agent’s scientific workflow without trusting a model to invent biology. Genomic pipelines can report success…

AgentDesk is a front desk for AI agents that any website adds with one line: app.use(agentdesk()), plus a MongoDB URI and an LLM key. It starts knowing…

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…

Recoup is the SLA-credit claims agent that remembers every win and loss in MongoDB, and whose next claim is better because of its last one. It reads GitHub's…

Laboratory automation is advancing, but scientific research still faces hard limits: available space and robots, costly reagents, and the time biological…

Hydra is an autonomous, self-healing AI agent framework designed to eliminate schema drift and breaking API contracts in production. When an agent encounters…

Forge: an AI agent that rebuilds its own harness, and can't cheat to do it. Forge takes a small, frozen model (gpt-4.1-nano) and lets an agent improve…

A self-improving AI agent that transforms GPU profiling data, technical research, and source code into ranked, actionable optimization tasks for open-source…

Precedent is a governance harness that lets company AI agents act, learn and be stopped, safely. Agents in companies fail in two ways: they do things they…

Every robot in a real building eventually meets a fire door, a wet floor or a crowd it cannot handle, and a human takes over. Today that takeover is wasted…

Your agent can rewrite its own harness. It can never grade its own homework.

Agent Recall is a persistent memory layer for AI agents that helps them retain, retrieve, and reuse knowledge from past experiences across tasks and…

https://shed-molt.fly.dev Shed is a recursive harness for agents. It observes which tools, rules, and memories a task uses, removes unused items while…

Recursive style agents drift toward generic, catalog-clean outfits and have no way to know when they’ve left a brand’s taste. We turned street-style photos…

LifeGuide is an AI-powered life navigation system that remembers important events, goals, preferences, decisions, and outcomes over time. When a user faces an…

Strata: an AI research agent that can’t forget its failures An AI agent that forgets its failures will report its luck as discovery. When an agent tests many…

Fit Gate is a daily outfit stylist built as a self-rewriting agent harness. Pick a mood and occasion, and it styles looks from your real closet using…

hx is a terminal coding agent that improves its own working environment from its own failures. Coding agents keep repeating the same mistakes across sessions…

Airlock serves as a self-improving controller for delegating and evaluating subagents within coding harnesses. In our own experiences, we've found that the…

We help personal bots create successful bots without having to test it.

Cash-heavy businesses (I'm involved with one: Thillens) reconcile deposits, invoices, and billing lines by hand every day, and the same fixes repeat for years…

We developed a self evolving harness using evolutionary algorithms to play Minecraft

HarnessEvolve is a returns operations simulation that shows how an AI agent can improve from its mistakes. It compares a fixed agent with an adaptive harness…

debugging web and code bases

Warehouse robots can waste time queuing at crowded lifts while faster routes go unused. Integration is a broader industry hurdle: 42% of surveyed businesses…

Quarantine is a self-improving harness for an IT-operations agent. Every tool call goes through one MCP gateway that enforces a policy stored in MongoDB…

OwnerOrbit gives restaurant owners one place to delegate work, track business goals, and receive proof that actions were completed, starting with a simple…