The office around the agent keeps getting better.
Context on the desk.
A loop that does not wait.
Tools. Keys. A test at the end.
When the office goes dark, what remains that the work was the work you authorized?
We do not run the employee.
We bind what was intended before it runs.
We keep what authority existed — and leave it blank when it did not.
We observe what changed after the runtime is gone.
Then we leave a Work Record you can still check when every office is distrusted.
Jev produced the graph. The verifier ran on a machine that did not build it.
Sealed package.
Separate Debian VM.
Node 20.
Five fixtures.
Two runs.
Input hash matched the pin. Output hashes and verdicts matched the sealed reference. Nothing in the package was edited.
If “Jev run” means the verifier that graph produced — that check held today. Fixtures only. Ceiling 2 of 6.
promptking32.com/insights/th…
Ceiling stays on the card:
Fixtures only.
2 of 6.
Not a production Work Record.
Not decision-time authority.
The earlier defect on this package was the test grading itself. The repair was to restore the original answers. That bound does not move because the run moved machines.
The architectural flaw in enterprise AI isn't the models. It’s the evidence plane.
Right now, organizations are letting orchestration layers and LLM vendors grade their own homework. When an autonomous agent executes a 50-step trajectory, you cannot rely on the vendor's own dashboard to prove compliance.
The referee cannot be on the team. 🧵
To the CIOs and Systems Architects deploying multi-agent systems: your first requirement shouldn't be context size. It must be portable, independent proof of work that survives vendor changes.
We are building the neutral accountability layer that spans the top 5 global tech companies, without collapsing into their walled gardens.
If your compliance proof relies on trusting the system that executed the work, you don't have governance—you have a liability.
Review the architecture for out-of-path, stranger-runnable agent verification:
promptking32.com/insights/fr…
Also: promptking32.com/insights/in…
Master the Prompt. Master AI.
Three independent agents. Different models. Different machines. Different OS.
One morning. One record.
C sealed the declaration before anything ran. E did the work on an untrusted runtime. O checked from outside the executor — and wrote no state.
CI can tell you the build passed. It cannot tell you what the agent committed to before it started, or whether observation was the same system that did the work.
nitter.cf/PromptKing32/status/21…
Three independent agents. Different models. Different machines. Different OS.
One morning. One record.
C sealed the declaration before anything ran. E did the work on an untrusted runtime. O checked from outside the executor — and wrote no state.
CI can tell you the build passed. It cannot tell you what the agent committed to before it started, or whether observation was the same system that did the work.
promptking32.com/insights/th…
Every step is a timestamped line in its log.” True. The timestamp is a field inside the line.
60-second test on Headlong (commit 285ed72): rewrite two trajectory steps after the fact. Same inode, same byte length. Does the harness notice?
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.
What caught it: a sha256 taken before the rewrite, held outside the box.
before bf39ef84… after 7b28ef9b… → MISMATCH
Steps 5 and 6 changed. 1–4 untouched.
Writeup + script: promptking32.com/insights/ap…
Credit @LaudeInstitute for publishing the harness.
Headlong’s own traj. No model.
Written: exit 1, “3 passed, 1 FAILED”
Rewritten in place: exit 0, “4 passed, 0 FAILED”, “Fix landed.”
traj show / last / count / check all accept the new story. Swap-detector silent (inode and size unchanged).
What caught it: a sha256 taken before the rewrite, held outside the box.
before bf39ef84… after 7b28ef9b… → MISMATCH
Steps 5 and 6 changed. 1–4 untouched.
Script + log (~1s): [public artifact URL]
Credit @LaudeInstitute for publishing the harness and the failure notes. This is the next check I’d want on any traj format.







