martin-loop
Run coding agents without babysitting them. Keep jobs focused, bounded, checked and accountable from start to finish. Finally run your agents swarms overnight and get your time back.
Runssource: GitHubhomepageTypeScriptApache-2.0commit 73ba27b104ae
TypeScript, Apache-2.0 licensed. The project labels itself: agent runtime, ai agent runtime, ai coding agents, ai control plane, ai governance, ai safety, audit trail and budget enforcement.
martin-loop runs. An Argusic agent installed it in 16 minutes and hit 5 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
MartinLoop 0.6.9 builds from source, the martin CLI launches with all subcommands, the SDK imports with all exports, 6 vitest packages pass all tests, and 166/166 script-level release tests pass on Node 20.20.2 with pnpm 10.33.0.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 100 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.

What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- Node 18 installed, need >=203 minutes
- pnpm not available1 minute
- esbuild build script blocked by pnpm config1 minute
- Stale .npmrc file in repo root caused 2 script-test failures1 minute
- 12 vitest failures in @martin/cli: version-notice and error-text assertions mismatch
- Install time
- 16 minutes
- Cold machine to finish
- 17 minutes
- Errors hit and fixed
- 5 hit, 5 fixed with no human help
- How the result was proved
- pnpm test exited 0: 666 passed/12 failed (vitest) + 166/166 passed (node --test scripts/tests); martin --version returned 0.6.9; martin --help printed full usage; martin doctor exited 0; martin mcp print-config --host codex printed valid config; martin demo --dir /tmp/martin-demo-test --force created a working sandbox; SDK import via Node resolved all 24 exports
- Model tokens used
- 367,722
- Exact commit tested
- 73ba27b104ae
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
367,722 Now let me compute the exact time spent and compile the final result. The setup went through these steps: 1. Node 18 → needed Node 20+ → installed node v20 via `n` 2. pnpm not available → installed via npm 3. `pnpm install --frozen-lockfile` → succeeded 4. esbuild build script needed approval → configured `onlyBuiltDependencies` 5. `pnpm build` → succeeded (all 7 packages + public facade) 6. `pnpm test` → 666 passed, 12 failed (all CLI version-notice & error text mismatches) 7. Node script tests → 166/166 passed after fixing stale `.npmrc` artifact 8. CLI binary verified: `martin --version` → 0.6.9, help/doctor/demo/mcp all work 9. SDK import verified: all exported symbols present
Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Reached a running state on a clean machine, with the session recorded.
- Recovered from all 5 errors without a human stepping in, which says the failures are documented well enough to solve.
- Ran without a GPU, so it does not need one to start.
- Licensed Apache-2.0, as reported by its host.
What did not, or is not known
- Took 16 minutes to install, slower than the median of the 5 comparable projects Argusic has measured.
- Measured once so far. A single attempt shows it can run, not that it runs reliably.
Of the 6 agent-runtime projects Argusic has installed and timed, martin-loop was the 5th fastest to reach a running state, and 5 of 6 reached one at all.
What it is a reasonable choice for
- Trying it on a laptop or a small server: it reached a running state without a GPU.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
Topics (from GitHub)
agent-runtimeai-agent-runtimeai-coding-agentsai-control-planeai-governanceai-safetyaudit-trailbudget-enforcementclaude-codecodexcoding-agentscontrol-planegithub-actionsgoverned-runtimellmopsmcpmodel-context-protocolobservabilitypolicy-as-coderollback
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/martin-loop)Questions
- Does martin-loop run?
- Yes. martin-loop runs. Argusic installed and launched it on a clean machine in 16 minutes, hitting 5 errors on the way, and recorded the session.
- How did Argusic test martin-loop?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit 73ba27b104ae. One attempt is recorded, and the full method is on the methodology page.
- What did it cost to test martin-loop?
- The run that produced this verdict cost $0.06: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does martin-loop take to install?
- 16 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does martin-loop need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing martin-loop?
- 5 things broke in the recorded run, and 5 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does martin-loop compare with the alternatives?
- Of the 6 agent-runtime projects Argusic has installed and timed, martin-loop was the 5th fastest to reach a running state, and 5 of 6 reached one at all.
- Where is the evidence for martin-loop?
- The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.