Subbu Athikunte retweeted
Introducing Headlong, an open source microharness for persistent agents: self-guided agents that think continuously.
Most agent harnesses are reactive: you send a task, the agent completes it, and then it sits frozen until the next request. Cron jobs and heartbeats wake it up to run a checklist and put it back to sleep.
A Headlong agent is never asleep. It keeps generating thoughts about whatever it decides is interesting, in a self-guided loop inspired by human inner monologue. Your message doesn't start a session. It's one more observation that lands in the agent's thought stream, and the agent decides if and when to reply.
Headlong is built on the idea of persistent agency: continuous inner thought generation between external interactions. The agent sets its own interests and priorities, comes up with its own projects, and sometimes pings you unprompted with progress.
To keep our prototype as simple and small as possible, we implemented Headlong as a microharness: a complete agent harness in under 10K lines of Bash, organized as a handful of small executables. It includes a loop that generates the next thought, shellm (a recursive language model written in Bash), a trajectory stored as a DAG of jsonl files, and context as a projection of that trajectory.
We've been running one Headlong agent internally at Laude for several weeks. The whole team talks to it over Slack and Telegram, and every conversation lands in its single stream of thought. It works in its own fork of Headlong and we've pulled over 50 of its commits into main.
One night, with nobody talking to it, it went back to check whether a recall process it had built was actually wired into its mind, found that it wasn't, diagnosed and fixed the bug, and verified the fix end to end. 48 minutes, no human asked for the fix or was in the loop at any point. Every step is a timestamped line in its log.
Things broke too, and we wrote those up. Background thinking costs us $1 to $2 an hour, our agent stopped its own service three times by accident, and self-delegation died on day one. Details in the post.
One line installs everything and starts an agent. Use a dedicated sandbox and spend-capped API key; it runs real shell commands and thinks around the clock.
Headlong is research software, be careful!
curl -fsSL headlong.ai/install.sh | bash
Launch post: laude.org/updates/headlong-a…
Repo: github.com/laude-institute/h…
Headlong is a @LaudeInstitute / MIT collaboration.
Subbu Athikunte retweeted
Evening! We’ve gotten lots of great feedback on the new ChatGPT desktop app (which we didn't get totally quite right on the first try), and as a result, we've made some changes.
1/ ChatGPT conversation history and projects are now visible in the sidebar. Also, your Chat and Work history now sync across web, mobile, and desktop. Local tasks still stay on your computer.
2/ You can now easily switch between Chat and Work modes inside ChatGPT on desktop, which is now also consistent with how it shows on web and mobile.
3/ Nothing is changing for users on Codex mode. It's still the OG and best at what it does.
And overall we're continuing to fix paper cuts and improve performance, reliability, and efficiency.
Keep up the feedback, hope you like the updates!
I just replaced a bunch of Mac Cleaner apps with codex. Way better than the tools and you can really control what to keep what to delete including all those stale node modules and homebrew/docker junk.
86
I have been using Codex a little differently this week.
Earlier, I would ask one agent to do a task, then I would review the output myself.
- This time, I kept one thread as the orchestrator.
- Then I used separate threads for the work.
- One thread made the PR.
- Another thread reviewed the PR.
- The first thread fixed the review comments.
- The review thread checked it again.
- Only after that did the orchestrator move to the next piece.
this is "loop engineering" within codex.
The prompt is only the start.
The useful part is the loop around the work.
Who owns the task?
Who reviews it?
What happens when the review finds a problem?
When is the work done?
When should the human step in?
This is the part that makes multi-agent work useful for engineering.
Starting agents is easy.
Closing the loop is the hard part.
80
I managed to get the loop working from inside codex. It created 7 threads and acted as an orchestrator to manage and reconcile PRs at the end. pure magic
If you're curious how I managed to do over $20,000 in inference on the last 48 hours, here's a video all about it.
Spoiler: loops are really powerful
76
How do you setup a large number of MCP servers to test a MCP gateway? MCP Zoo. I just built a mock server that mimics different properties of these servers. To make it fun to use I just got gpt-image-2 to generate a nice image.
Day 10: Enterprise MCP Gateway development
50
Using Codex to drive Claude. Claude acts as UI engineer and Codex the tester.
Day 9: Enterprise MCP Gateway development
48
Day 8 - Designing Enterprise MCP Gateway
I have been using OpenAI's symphony to develop the mcp gateway. This is the current session. Codex app server spinning up 5 codex sessions in parallel to implement 5 todo's from the Linear board.
51
This is really good with Indian languages. "create the front page of a Hindi newspaper dated May 13, 2023" picked up news from that day and created an image. Very impressive.
🤖 Made with AI
Replying to @OpenAI
Stronger Across Languages
ChatGPT Images 2.0 can produce images with non-English text that’s not only rendered correctly but with language that flows coherently.
This makes the model more globally useful and helps people create visuals that work in the languages they actually use.
3
184
Finally hit the limits of a $200 pro plan in codex. 23 hours of continuos run. Multiple auto compacts. Surprisingly, codex doesn't stop mid session even when its out of messages.
1
52
Subbu Athikunte retweeted
"Using coding agents well is taking every inch of my 25 years of experience as a software engineer, and it is mentally exhausting.
I can fire up four agents in parallel and have them work on four different problems, and by 11am I am wiped out for the day.
There is a limit on human cognition. Even if you're not reviewing everything they're doing, how much you can hold in your head at one time. There's a sort of personal skill that we have to learn, which is finding our new limits. What is a responsible way for us to not burn out, and for us to use the time that we have?" @simonw
"Using coding agents well is taking every inch of my 25 years of experience as a software engineer."
Simon Willison (@simonw) is one of the most prolific independent software engineers and most trusted voices on how AI is changing the craft of building software. He co-created Django, coined the term "prompt injection," and popularized the terms "agentic engineering" and "AI slop."
In our in-depth conversation, we discuss:
🔸 Why November 2025 was an inflection point
🔸 The "dark factory" pattern
🔸 Why mid-career engineers (not juniors) are the most at risk right now
🔸 Three agentic engineering patterns he uses daily: red/green TDD, thin templates, hoarding
🔸 Why he writes 95% of his code from his phone while walking the dog
🔸 Why he thinks we're headed for an AI Challenger disaster
🔸 How a pelican riding a bicycle became the unofficial benchmark for AI model quality
Listen now 👇
youtu.be/wc8FBhQtdsA
Subbu Athikunte retweeted
We made LLMs speak Tulu, a language with only 2 million speakers.
It wasn't easy because LLMs kept confusing it with Kannada, but we discovered negative constraints really help.




































