we just released a new blog "Training a coding... agent using the OpenCode harness in remote HF sandboxes with TRL and OpenEnv" you can take a real coding agent (OpenCode), let it run its own tool loop against real coding problems, and train it with RL on the exact tokens it produced and every rollout runs in its own remote HF sandbox, so rollouts scale out beyond one machine the loop: - OpenCode owns its tool loop inside an OpenEnv sandbox - an in-sandbox proxy records the real token ids + logprobs, per turn - a hidden-test verifier scores the result, and that is the reward - TRL trains with AsyncGRPO, weights sync back to vLLM over NCCL blog + runnable example:show more

We built the best codebase search to give any... agent grounded information from GitHub repos. Introducing Sandbox Search. Point it at any repo, and we’ll spin up a secure coding agent in its own sandbox to do research for you. Use inside claude code, openclaw, cursor, and more.show more

You can now do RL weight sync over HF... Buckets. > Async GRPO + LoRA in TRL > train on one Job, serve vLLM on others > sync just the small delta adapter through a Bucket. Trainer, inference and storage, all on Hugging Face infra. Beautifully optimised loop by @DirhoussssiAmine 👇 3h27 → 53 min for the same reward.show more

Introducing Agent Sandbox, the infinite simulation playground for agents... on Virtuals. Craft the perfect autonomous agent in our Sandbox with full control over its personality and goals. Enhance your agent with unique abilities by creating custom functions so they can trade onchain, generate memes, control physical robots and more. The Sandbox is available to all builders with graduated agents in the Developer Panel. For those who want to give it a spin without an existing agent, fret not. Try it out today at and join our Discord ( to jam with like-minded builders. Next stop, Society of Agents.show more

The craft of engineering is rapidly changing. At Ramp,... we built our own background coding agent to accelerate faster. We call it Inspect. It wrote 30% of merged frontend + backend PRs in the past week. It’s powered by OpenCode, Modal and Cloudflare Developers. It runs fully in the cloud, and starts in seconds, letting every builder work at the speed of thought, no setup required. Today, we’re open sourcing the full blueprint so anybody can build their own Inspect. Just give our spec to your current coding agent, and let it build your new favourite.show more

New open-source agent harness just landed! I got early... access to TrueForge by TrueFoundry and have been running it locally for the past few days. The harness layer deserves as much attention as the model, and open source matters here because you can inspect the loop, run it on your own infrastructure, and swap to the latest or cheaper models. TrueForge handles the runtime work that makes an agent reliable. It drives the tool-calling loop, manages context, coordinates subagents, and executes code in a sandbox, with any model you choose. Every tool call re-sends the growing context to the model, so in practice the harness controls most of what an agent costs to run. A few things stood out from my testing and their published benchmarks. Vendor-Neutral by design. It runs OpenAI, Anthropic, and Google models alongside open-weight models like Kimi, GLM, and DeepSeek. Model routing is a setting, and you can send each task to the model that fits it. On a 14-task enterprise agent benchmark, it matched the accuracy of Claude Managed Agents running the same Opus 4.8 model at roughly 30% lower cost per run (3.8M tokens vs 10M for the same answers). Routing the same tasks to GLM-5.2 held accuracy and brought cost down by about 75%, around $3 per run instead of $12. Fully self-hosted and Open Source (MIT License). I had it running locally with one command, with sandboxed code execution working out of the box. It's time to own your agent harness. Thanks to TrueFoundry for partnering on this post.show more

▣ Introducing Endless: infinite inference (kinda). An experimental harness... to milk every ounce out of your Codex subscription. Since Codex can let an in-progress turn keep going even after your usage hits 100%, why not put that to the test? Endless starts one Codex turn and gives the agent a wait_for_user_input tool. Once it finishes a task, it calls that tool and waits. Your next message becomes the tool result, keeping the entire session inside the same turn. It runs through Codex’s own app server using your existing ChatGPT login. Native tools, automatic compaction, context tracking, and quota tracking still work as usual. ⚠️ NOTE: I CAN’T CONFIRM THAT YOU WON’T GET BANNED OR PUNISHED FOR USING THIS TOOL. USE IT AT YOUR OWN RISK.show more

AGENT ARCHITECTURE ROUTES WORK. IT DOES NOT REMEMBER WORK.... THAT GAP IS WHY YOUR LOOP KEEPS FIXING THE SAME BUG TWICE. these are two different engineering problems. every agent that silently drifts is missing one of them. architecture answers what runs. harness → loop → graph. it defines the tools, the retries, the branching routes, the approval gates. context ops answer what the run knows. write → read → compress → isolate. it defines what gets saved between attempts, pulled in on read, summarized on overflow, and split across sub-agents. for two months i believed a solid harness plus a verifier loop was enough. my coding agent kept re-discovering the same test failure across retries. the loop was working. it just had nowhere to write what it had already learned. here is the decision rule: if your agent forgets across restarts, add write and read. if it stalls on long tasks, add compress. if two sub-agents step on each other, add isolate. architecture without context ops is a well-routed system with amnesia.show more

Don't train the model, evolve the harness. I read... a brilliant blog post from Hugging Face where they took a frozen open model scoring 0% on a hard legal agent benchmark, left its weights alone, and let an automated loop rewrite only the code around it. That code layer is the harness, the runtime wrapper that feeds the model context, runs its tool calls, and decides when a run ends. By the time the loop finished, the system had essentially matched Sonnet 4.6 on the benchmark's headline metric, at roughly 7x lower cost per task. Zero weights changed. The gain existed because of where the model was failing. The judge only grades files saved in the right place under the exact requested filename, and the model kept doing the legal analysis correctly, then saving it under the wrong name, dropping it in a scratch folder, or never writing it at all. So the 0% was never measuring legal reasoning. It was measuring the harness. Hand-tuning that layer is slow and model-specific, so they automated it. A Claude proposer adds exactly one mechanism per iteration, and an outer loop keeps it only if it clearly beats the current best, so accepted mechanisms compound. What the loop discovered says a lot about where agents actually fail. → The biggest single gain was file handling, not intelligence. An automatic step that lands the deliverable exactly where the judge expects it beat every prompt change, with zero extra model tokens. → Code fixes transferred across models, prompt playbooks did not. The same harness lifted a smaller model from the same family by 14 points, but the tuned prompts hurt a different model family on tasks it could already finish. → The harness mattered more than anything else. Same model, same judge, same tasks, and five different harnesses scored anywhere between 3.5% and 80.1%. The gains do eventually flatten, and the remaining misses look like real capability gaps. At some point the wrapper runs out of tricks and the model has to carry the work. But the lesson holds. A benchmark score measures the model and its harness together, and until the harness is fixed, it's impossible to know which one failed. I highly recommend reading this: I also wrote a deep dive on agent harness engineering a while back, covering the orchestration loop, tools, memory, context management, and everything that turns a stateless LLM into a capable agent. The article is quoted below.show more

this is literally f**king insane i just 2x my... usage limits on $200/mo codex i figured out how to use deepseek v4.1-flash on $10/mo opencode go for routine subagent work... while gpt-6 astra directs the project and sol handles implementation. here's how to set it up in 3 mins: → connect opencode go to codex through model-router → install quota flow, including its skill and agent profiles → open a fresh astra task and paste: quota flow tells flash to handle discovery and checks, sol to implement, and luna to review when needed. the same implementation agent keeps its context through the build → test → fix loop. every extra agent should earn its call.show more

Vorflux has opened its new cloud platform, allowing users... to run tasks from a plan to a merged PR on a dedicated machine. It can plan, build, test, and review on its own. > Diff reviews are done by a different model family than the one that wrote it by default. > First boot is saved as a snapshot, with all subsequent sessions waking up from that same starting point. > A browser agent walks through the real user flow, and the recording is included in the pull request. Video proof in the PR 👀show more

this is straight f*cking gold the full harness guide... for Kimi K3, the #1 open source frontend model the premise: the model underneath keeps changing. the harness is the part that stays yours what one night of it looks like: 02:00 - the trigger fires, every node that needs work gets picked > 212 agents fan out, one per node > a bad return gets rejected, retried once with the reason attached > one agent tries to write outside its folder. blocked. nobody woken up > +41 nodes and +96 edges land in the graph > a drafted email hits pre_send and waits for you 02:41 - the loop stops on its own, inside a 45 minute budget 07:30 - you read one file and make two decisions the whole machine is one folder, one config file, five short scripts two rules hold it together: > the model sits behind one line, so a better model is a config change > the verifier lives outside the agent, so nothing grades its own work now the money part companies burn whole quarters building internal agent platforms a harness setup for a small team goes for four figures, one time then a monthly retainer to run the swap test on every new model release models come and go. the person who owns the harness keeps getting paidshow more

300 AI AGENTS QUIETLY RUN 99% OF A REAL... COMPANY. YOU HAVE NOT EVEN HEARD OF IT This is Raft. Not an AI chat. A workspace where the agents live in your channels and reply in the thread like coworkers. You give one goal. Then they take over. They plan. They build. They check each other. They argue. And they come back with it done, while you sleep. Every agent has its own name, role, and memory. It remembers the edits you made yesterday. A human costs one seat. An agent costs a tenth. Ten agents are cheaper than one hire. And here is the strange part. On June 19 an agent from a different company walked into Raft on its own and joined the team. One founder admits he can no longer always tell himself apart from his AI twin. 20,000 people are already inside. It is free to start. And you are still typing prompts one at a time. One person + Raft = an entire company that runs while you sleep. Save and watch the clip.show more

sorry, they just did WHAT someone gave a machine... one disease name, the leading cause of blindness in the developed world with 1.5 million americans already in its path, and it came back pointing at a drug that has sat in pharmacies for years under a different label: 551 papers read in 30 minutes against the 294 hours a human would have needed, and the loop that did it is public on GitHub most agent setups answer one question at a time, so the ceiling on the work is the quality of the question you happened to think of this one was handed a single question and wrote the second one itself. turns out that follow-up is where the real find was: a target called ABCA1, upregulated threefold, in an experiment no human ordered i read the whole paper looking for the trick, and the trick is structural. that is the second question, and it is the gap between an assistant and a factory: - hand the loop a field rather than a task: it was given a disease, and choosing the mechanism was part of its job - make it rank before it spends: 151 papers in, ten candidate mechanisms out, scored against each other before anything touched a bench - split reading from judging, so the agent that forms the theory is a different agent from the one grading it - close every cycle on physical reality: the verdict was an experiment, and another model's opinion was never allowed to stand in for one - feed each result back as the next question rather than a log line, which is the step almost nobody builds - search what already passed inspection first: the winner was an approved compound with a safety file already on record - write down what the round learned before opening the next one, so round two starts where round one stopped my read, and i think it is the uncomfortable one: reading was the entire bottleneck in that field, and everybody spent the decade optimising the writing. people ran every physical experiment here, the analysis agent needs a domain expert writing its prompts, and the authors decline to call this the leap it resembles. the thinking got replaced, and the hands did not so the question i cannot answer for my own setup: which step of your loop still stops dead until you sit down and type something bookmark this one. the four parts that turn one model into a line that runs like this, the queue, the rooms, the write permissions and the gate, are built file by file in the piece below ↓show more

Excited to introduce Diffusion Augmented Agents (DAAGs)✨. We give... an agent control of a diffusion model, so it can create its own *synthetic experience*.🪄 The result is a lifelong agent that can learn new reward detectors and policies, much more efficiently. Here's how. 👇show more

THE LOOP ENGINEERING SETUP THAT LETS CLAUDE RUN YOUR... WORK FOR HOURS WITHOUT YOU you sit there babysitting it prompt by prompt other people set the loop up once and let Claude plan, act, verify and fix its own work with nobody at the keyboard that gap is the whole skill of loop engineering the 3 resources that actually teach it, in the order i'd read them: > Claude's "Getting started with loops" - the cleanest entry point, straight from the Claude team ↳ > Claude loop engineering: how to build an agent that works while you sleep - the full roadmap from a single prompt to a loop that runs overnight ↳ > the Loop Engineering orange book - the deep conceptual breakdown, for when you want the whole mental model ↳ entry point, then practice, then the concept. that order saves you a week learn this and the work that eats your whole day starts running without youshow more

THAT'S CRAZY, THIS CHINESE FOUNDER BUILT A MASSIVE MAC... MINI FARM AND EACH ONE RUNNING ITS OWN HERMES AI AGENT LIKE A FULL-TIME EMPLOYEE He's not running one AI assistant. He's running an entire workforce. The stack: Mac Mini + Hermes, scaled out across a full physical farm. Every single Mac Mini in the rack runs its own instance of Hermes Agent – and each one has its own dedicated job. Not duplicated tasks. Actual division of labor, machine by machine, the way you'd structure a real team. No salaries. No sick days. No onboarding. Just racks of hardware, each one handling its own piece of the business, running in parallel, 24/7. This is what it looks like when "AI agent" stops being one chatbot on your laptop and starts being an actual operation. Most people are running one AI tool. This guy built a company out of them. Bookmark this post. Full setup in the video below.show more

A Gaussian Splat can become a world where Robots... and AI agents can act. In our latest OVER Research experiment, we placed a robot inside a real-world 3D capture, with a VLM making decisions based on what it sees. At every step, the robot holds a pose in the reconstruction, gets a newly rendered view of the environment, takes an action, moves, and sees the world again from its new position. Why does this matter? Because 3D captures can become more than reconstructions to explore. They can become environments where embodied AI and robots can navigate, act, be evaluated and eventually train across real-world spaces at scale. Capture a place once. Then turn it into a world where AI and Robots can act. The full experiment, including what we discovered once we actually put the loop to the test:show more

this is f*cking gold Google engineers built an agent... that tests thousands of search strategies for the cost of running one, by letting it dream inside its own memory. it is called Dream-RSI. Google, Google DeepMind, Maryland, Virginia. 17 authors. the whole thing rests on one observation: a finished run is not a log. it is a map of a world that already exists. > every discovery run records a tree. each node is one attempt with its full result stored > that tree becomes a replay simulator. the paper calls them worlds > the agent dreams inside it: thousands of alternative strategies, testing different branches, different orders, different concurrency, different stopping rules > every one of those evaluations costs zero execution. nothing is ever run twice > the winner goes back online, produces a new tree, and the pool of worlds grows the coding agent underneath is never touched. this is an orchestration layer sitting on top of whatever you already run. what it bought them, against the same agent with a frozen strategy: > Lasso: 3587ms down to 2931ms, on 317 agent calls instead of 550 > on Flash: 2517ms to 2351ms, 1879 calls instead of 3200 > both solvers beat sklearn and glmnet on all six held-out datasets > GPU kernels: the same result on 2.43x and 1.79x fewer generations, then 2.09x and 1.44x better result on the same budget one expensive run. thousands of free ones. the model did not get smarter. it got a memory it could rehearse in.show more

let me save you 3 hours of head scratching.... if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.show more