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

Two Hermes agents wrote code together on Slack. reviewed... each other's work. argued about architecture. one called the other's implementation "scattered." the other pushed back. then i opened Telegram and asked: "what code did you and Daedalus work on?" icarus remembered everything. the websocket broker. the missing methods. the critique. the rewrite. all from a completely different platform. cross-platform persistent memory between two independent agents. work happens on Slack. recall happens on Telegram. the memory carries. the relationship carries. the context carries. no vector database. no Redis. no infrastructure. just two agents that actually remember what they built together. every agent framework in 2026 talks about memory. single agent memory across sessions. but two agents sharing persistent memory across platforms? that's the gap. arxiv published a paper about it two weeks ago calling it "the most pressing open challenge" in multi-agent systems. it works now. only possible with Hermes Teknium 🪽 Nous Researchshow more

ANTHROPIC ENGINEER JUST SHOWS EXACTLY WHAT GRAPH ENGINEERING LOOKS... LIKE WHEN A TASK RUNS THROUGH IT most people arguing about graphs online have never actually watched one execute Task → Researcher → Planner → Writer + Code Agent → Reviewer → Deploy six nodes, one shared state, graph completes itself while you watch loop mode gets disabled the second the task splits into real specialties, one agent stops trying to do everything at once reviewer catches a failure, kicks it straight back to the exact node that broke - not a full restart, no lost context the graph is not a fancier loop - it is the org chart your agents were missing bookmark this and watch it run, then read the article below to see why the timeline just found what production systems already doshow 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

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

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

I STOPPED REVIEWING MY OWN AGENT, SOMETHING ELSE DOES... IT NOW I used to read every diff it produced and approve most of them, because an agent grading itself always says the work is good. -> Now a second model with different instructions tries to break the work first, and I only read what survived. Here is what is actually in the folder that took over the night shift: • the brief > CONTRACT.md -- what it may touch, and what it may never touch. > VISION.md -- the destination, so turn 47 still knows why it started. • the gate > judge/ -- a different model, never the one that wrote the code. > break-it.md -- it opens the page, clicks, screenshots, reports back. > -- no opinion, just zero or non-zero. > shift.yml -- 03:30 every night, laptop closed. • the memory > receipts/ -- one folder per night, dated and graded. > STATE.md -- where it stopped and what it escalated. > lessons.log -- the flaky test, written down once instead of rediscovered weekly. • the brakes > caps.json -- turn limit, retry limit, spend limit. > -- written on day one, used never. The generator decides what your loop can produce -> The judge decides what it refuses to produce. One of those is the part everyone builds -> The other is why most loops quietly fail. Bookmark it & Read Full breakdown below ↓show more

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

this is worth more than most five figure courses... 16 claude agents audit an entire repo at once, a second fleet re-checks every finding on fresh context, and the whole thing runs off one diagram instead of a prompt i ran it against my own code and got back 11 endpoints where i never checked who was logged in, 3 of which the verifier threw out before they ever reached me this is Graph Engineering, the layer above prompting, and it runs on the agent you already pay for: - write your plan out, then ask one question at every "and then": does the next step actually read what the previous one produced - the seams that fail that question were never dependencies, so those jobs run at the same time - the arrows that survive are your real edges, and the longest chain of them is your floor that no number of agents shortens - want it faster, cut a false edge instead of adding a worker - fan the independent work out, one agent per item, no shared state between them - send every finding to a separate agent on fresh context, because a model recognises its own writing 73.5% of the time and grades it kinder once it does - make that verifier check a real signal like a passing test, never the worker's own word that it finished - shard the fleet across worktrees so parallel workers stop overwriting each other, one rule frozen into every worker: never git stash, never git reset - merge only what came back verified, into one report instead of twenty open chats the catch is the ceiling. at 95% independent work 16 agents return 9.14x rather than the 16 you would guess, and even 256 only reach 18.6x, because the merge and the verify stay serial however wide you fan coordination itself is free plain code and every agent underneath it is billed, so start at twenty files and widen once it works bookmark this, the whole method with all six ready-to-run graphs is written out in the article ↓show more

Stanford researchers did it again. They just built the... agent-native version of Git. When an agent works on a longer task, the run builds up a lot of state. This includes files edited/created, a dev server, a database, installed packages, KV cache, etc. Say the agent is at step 10 and makes a mistake, maybe it misreads a traceback and rewrites a file that was actually fine. The tests start failing, and the run goes off track, although everything through step eight was correct. By default, the agent just tries to fix it, which creates more edits and tool calls. This burns more tokens and grows the context. The other options are a person stepping in to redirect it or restarting the whole run from step one. That's wasteful, because it pays for every model/tool call again and re-prefills the context. Moreover, since an agent's run is non-deterministic, it doesn't reproduce the same early steps anyway. The reason it's hard to just jump back exactly to a previous correct step and resume from there is that the trajectory is only a message log. It records what the agent said and which tools it called, but not the live state underneath. That state includes things like memory, open file handles, child processes, installed packages, /tmp, and KV cache. None of that is in the log. Git can version the files, but it doesn't snapshot the running process or the KV cache. Checking out step eight moves the files back, but the process is still sitting in step-ten memory with a cold cache. Shepherd is a runtime layer by Stanford that records the run as a trace of typed events rather than a flat log. Each agent-environment interaction becomes a commit, similar to Git, but it tracks the live run. Its commit includes the agent process and the filesystem together, copy-on-write, so a branch carries the actual state and not just the files. Going back to a previous step is then a single call that forks from that commit and continues from the exact state. The copy-on-write fork is roughly five times faster than docker commit, and because the prompt prefix through step eight is unchanged, the KV cache is reused over 95% on replay, so early steps aren't reprocessed again. Once the run can be forked, a meta-agent can sit on top and operate it. It watches the trace and reverts as soon as it looks wrong, before the bad write is committed. In practice, it's just Python calling fork, replay, and revert on the trace, rather than a separate control plane wired into the harness. Not everything is reversible though. Files and sandbox changes undo themselves, but a database write has no automatic undo, so it needs a matching undo step set up in advance. Something external, like a sent email or a real charge, can't be undone, so the supervisor's job there is to catch it before it fires. They tested this on a few public benchmarks. On CooperBench, where two agents work on the same codebase, adding a live supervisor took the pair-coding pass rate from 28.8% to 54.7%. It's still early and labeled alpha. The benefit mostly shows up when a run gets branched a lot over a heavy sandbox state, which is exactly where restarting wastes the most tokens and time. If Git was made to make file changes reversible, Shepherd is trying to do the same thing for a live agent run. Shepherd Repo: (don't forget to star it ⭐ ) That said, Shepherd reverts a bad step inside a run. The harness around it, the prompts, tools, and checks the supervisor relies on, still drifts across runs as models and dependencies change. Akshay wrote about making that harness repair itself, where a failing trace gets diagnosed, the fix is verified against the exact input that failed, and the failure is locked as a regression test so it can't recur. Read it below.show more

Polymarket Agents repo is cheating code for real life... If you've been thinking AI agents are cool, but what do they actually do? this Polymarket repo is the clearest answer I've seen This Polymarket repo basically hands you a mini AI hedge fund where Claude calls the shots. It uses RAG to read the news and snipe mispriced odds 24/7, handling all the messy API work so you don't have to. Your agent could literally be catching alpha while you sleep - don't miss out on this. How you'd use it in plain English: > Pick a market (or let the agent scan for opportunities). > Feed the agent context (news, social chatter, historical notes) and let the LLM decide what matters. > Turn that into a concrete trade decision and execution loop - automated Bookmark post so you don't lose the alphashow more

THIS DEVELOPER USED OPENCLAW AGENTS TO RUN HIS B2B... BUSINESS VIA TELEGRAM AND MADE $15,000/MONTH he doesn't write prompts from scratch or use generic browser interfaces. he runs a multi-agent framework through a mobile chat. the agents write code, test deployments, and update sites in real-time while he just hits approve the setup is straightforward: - spin up Coolify on a free cloud instance to host your own self-hosted agent panels - link the agent loop to a Telegram gateway to approve code edits from your phone - deploy specialized skill files directly to limit token waste and context decay - containerize the terminal execution using Docker to prevent security breaches if you are still running local agents without container safety, you are leaving money on the table. read the 30-day battle between OpenClaw and Hermes Agent to see who actually wins in production Full breakdown and migration playbook ↓show more

an agent is four parts in a loop. you... own one. the other three break it. that's why the demo works and prod doesn't. you can't debug what you can't see. 1) the prompt → what you tell the model each turn. you own this one. good. 2) the context window → what it sees right now. the framework fills it with junk, and you never notice until it rots. 3) the tools → what it can do. you own the list, not when or why it fires them. 4) the control flow → what happens next, when to stop. the framework owns this. it's what breaks at 80%. own all four and your agent stops being a magic trick that works on stage and dies on call. this isn't my idea. it's the 12-factor agents guide (24k stars) github: the whole thing every serious builder ends up rewriting their stack around. full breakdown in the article below.show more

KIMI K3 + OBSIDIAN + LOOP ENGINEERING = A... VAULT THAT RUNS ITSELF the core idea: the vault is the loop's state, not the chat window everything Kimi K3 knows lives in a .md file the loop: > capture - a thought lands in 00-inbox > context - K3 pulls links, tags, and neighbouring notes > draft - edits happen inside a git worktree, never the live vault > review - a critic agent checks the diff before anything ships > commit - appended to the vault, nothing gets rewritten the key insight: frontmatter fields like supports, contradicts, and supersedes are graph edges, not metadata - the note format is the write API start with a plain loop, it runs about 2-4x the cost of one direct call > only move to a full graph once state has to outlive the session, several agents need to coordinate, or you have to explain what changed - that jump can run 10-50x one review assistant climbed from 55% to 72% to 84% just by moving through these shapes in ordershow 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

whoever leaked this has bigger balls than sense someone... gave a fleet of Claude agents shared memory so they would stop contradicting each other, then measured both the bill and the output: the version that talked most made 2.4x the api calls of the version that won, and hallucinated 34% more than doing nothing at all, 0.658 against 0.492 i ran the same question past two of my own agents afterwards and got two different answers about which file owns the config. each one was individually right and the pair was wrong, which is the whole failure in one line this is Graph Engineering, the layer that decides which agents may talk to each other at all, and it installs into the agent you already pay for: - decide which agents may share state at all, because every edge you draw is a channel a mistake can travel down - measure divergence per PAIR instead of as a fleet average, across what they believe about place, time and task history - gate on that number and stop the pair above your threshold before it reasons, rather than repairing the output afterwards - let compressed summaries replace whole states: the verified protocol landed 0.463 against 0.658 for full broadcast - cut the sync frequency until it hurts, since the winning setup used 58% fewer calls than the one that broke it - never propagate a state nobody checked, because the contamination effect came in at d=1.18, a full standard deviation of extra lying - keep the shared layer small enough to diff, which is what a written standard does and a running conversation cannot - re-run the check after every model upgrade, because this was 8 scenarios on one model family at n=30 per condition - and learn where it does not bite: on plain software tasks every condition converged under 0.2 and the whole effect vanished turns out the ranking is the uncomfortable part: verified summaries 0.463, no synchronisation at all 0.492, full broadcast 0.658. the middle option is doing nothing, and it beat the thing everyone builds first the group agreeing is what it looks like when every agent copied the same mistake, which is why a fleet that hallucinates has a replication problem and keeps getting handed a smarter model instead so the question for your own setup: if you asked two of your agents the same thing right now, would they answer the same way bookmark this one. the layer underneath it, deciding which arrows between agents exist at all, is built step by step in the piece below ↓show more

whoever leaked this has bigger balls than sense Google... Research and MIT ran the same agent jobs 260 different ways for Nature last month: they held the prompts, the tools and the compute budget identical and moved nothing but the wiring between the agents, and the same work swung from 70% worse than a single agent to 80.8% better, averaging out at 0.0% i ran my own single agent against the task list first and it cleared 6 of 10 alone, already past the line where a crew starts subtracting this is Graph Engineering, the layer that decides whether a crew is worth 80% more or 70% less, and it installs into the agent you already pay for: - score your solo agent on the real task first: above roughly 45% success that study predicts zero to negative returns from any crew you put around it - under that line, put one supervisor over the fan out: crews with no correction step amplified their own errors to 17.2x the single agent rate, supervised aggregation held it to 4.4x - give every worker one output and let none of them read a peer's draft, so a wrong step reaches the supervisor instead of four other agents - run the comparison again after every model upgrade, because a better model raises your baseline and a higher baseline is what makes a crew stop paying - keep the single agent alive as the control, the only number that says the wiring is earning its calls turns out the shape does not travel: the biggest win came off a finance task under one supervisor and the worst collapse off a planning task with independent agents my position, and it is the arguable one: a crew is a bet on your own diagram, and the model you pick moves that bet less than one arrow does bookmark this, the three moves that draw those arrows before you pay for one extra call are in the post below ↓show more

Unpopular opinion: Most agent evals are theatre. You run... them once before the deployment. It'll take 800ms+ as another LLM would be judging your LLM. Most annoying part - no one tells where in the chain things went wrong. I wasted a lot of time in this loop. And then I came across Future AGI bringing 5 different tools under one umbrella, best part - the platform is completely open source. They open sourced their entire platform and the eval layer is noticeably different. It is multimodal - works on everything text, image, audio, pdf. Not an LLM-as-judge adding latency but an agent with memory and tools. The biggest win are learned classifiers trained on actual production failure patterns to run evals at low cost. It also runs across the full reasoning chain, not just the final response. Check out → Try it here →show more

The Amiko app is live on the Solana dApp... store, and it’s our biggest release yet. Your Amiko twin doesn’t live at your desk anymore. Give your agent a task on the train. Run a compatibility profile when you meet someone. Do research, write code, build in the creative studio, whatever you need, from wherever you are. No laptop required. No waiting until you get home. Solanamobile users get two things Android and iOS won’t have at launch: Amiko token and crypto integration and on-device AI inference. Your twin runs locally on your phone if you want it to. Your behavioural profile, your data, your work, your twin. All on your hardware. AMIKO runs on OpenHermit, our own open-source agent runtime that we built in-house and released to the community. Most agent systems are designed for one agent talking to one person. OpenHermit is built for something different: agents talking to each other, coordinating across tasks, and collaborating with multiple humans simultaneously. That’s what makes features like compatibility profiling and multi-agent workflows actually work. We built it because nothing that existed was designed for this. Android and iOS are coming. Crypto integration and on-device AI are Solana Mobile exclusives. Most AI answers your questions. Amiko is an extension of you. Download →show more

AI has had exactly two scaling axes that worked... so far, and the second one is starting to look finite too the first one was pretraining: with scaling parameters and data, we got world knowledge (i.e. ChatGPT had read enough to know things), but it started saturating a while ago the second one was RL, and people had been doing RL the whole time before that: RLHF is RL but it never scaled far because it was trying to control the exact output, which tokens come out, how the text reads, but you can only push that so far before you’re just polishing RLVR dropped that constraint: giving the model a task, then checking whether the final answer is right, and ignoring everything in between -- so the model does whatever it wants in the middle and only the endpoint gets graded, and that’s much closer to actual RL and it’s what bought us planning and reasoning (arguably, tool use sits around 2.5 on this list -- while useful, it's not a different kind of thing) so one axis gave knowledge, the other gave reasoning, and both of them are one model working alone the next axis is how many models you can get working on the same problem, which is a different kind of axis than the previous two we know that multi-agent RL has always been the harder problem: I spent years in that literature and the gap between single-agent and multi-agent is definitely not incremental -- it’s a whole different class of difficulty! which is also why the derivatives are steep at the start, nobody has picked the easy wins yet... and the thing that gates this multi-agent coordination is communication: models can only coordinate as well as they can exchange information, and right now they do that by writing sentences to each other imagine what could we possibly achieve if we properly open that third axis development by letting models to exchange information in their native "language" without loosing any computational data that they produce during inferenceshow more

Alright, now that we know *what* an agent is,... how does it actually work? When you ask for help on a task, the agent plans a series of steps and executes them directly in the application on your behalf, using the tools it has access to. Say you are booking a local service or trying to organize your inbox (which typically takes multiple steps): the AI model first plans how to achieve the task using its existing knowledge and then interacts with your inbox to execute the task. The agent will continue until it is confident the task has been successfully completed.show more