Build-in-public log agent
Turn real development changes into a website update.
Useful development work often stays buried in commits. This workflow turns one concrete change into a readable website article with a record of what happened.
Running on our own company since Oct 31, 2026 · claw-world.app · Risk: Low. Everything on this page is public; there is nothing to sign in for unless you want us to run it for you.
Tasks and outputs
What it takes on
Collect changes
Fix the input range and reuse it on retry.
Write one update
Explain a concrete problem, the change and its limits.
Record the result
Export a summary for this log; verify an article before linking it.
Who it is for: Founders and small teams documenting their own product development.
How we use it: Running on demand since October 31, 2026. The first two reports cover October 19–25 and October 26–31. We aim for about two runs a week; social distribution is on hold.
Evidence before claims
01One useful topic per run; skip when there is nothing worth publishing.
02Do not describe a commit as a live feature without deployment evidence.
03Do not expose internal evidence or unpublished draft text.
04Social distribution stays on hold.
The rules the agent’s output is written to, from the file named above. This site prints no example output; what the agent has shipped is in its run log.
Workflow
How a run goes, step by step
Pick a step to see what goes in, what comes out, and where a person is involved.
Step 1 of 4
Collect
Read eligible changes since the saved position.
An explicit first range or the previous cursor
A fixed input range
Choose the first range.
Step 2 of 4
Draft
Choose one topic, verify its facts and write the website article.
Eligible development changes
A private draft, or a skipped or failed record
Review factual claims.
Step 3 of 4
Publish the website
Prepare the article, deploy when authorized and verify its live contents.
A reviewed article
A verified website URL
Approve website publication.
Step 4 of 4
Update the log
Export safe metadata and deploy the public summary file.
The local run ledger
This public log
Review the exported change before deployment.
Before you start
What you’ll need to bring
Company material
Product context and a clear audience.
Source material
Git history with an explicit starting boundary.
Platform account
Repository and website deployment access.
Run environment
Node.js, Git and an agent runner; no new service or API key.
Run log
Recent runs, as they happened
The full log is public and read-only. It holds every run on record, including the ones that stopped to wait for a person.
View the full run logGuides and files
Set it up yourself
The setup guide, and the files the agent is built from. You run it in your own environment, with your own accounts.
- 01
Runtime
- Node.js, Git and an agent runner that follows the workflow.
- 02
Browser setup
- No browser session is needed to draft an article.
- 03
Account setup
- Repository access; deployment access when publishing the website. No social account is needed.
- 04
Configuration
- Choose an explicit starting commit for the first run. Later runs resume from the saved position.
- Run about twice a week when useful changes are available.
- 05
First run
- Collect a range, inspect the evidence and save one website draft.
- Review the public summary, then deploy it to update this log. Publishing the article is a separate step.
build-in-public-log.md
Workflow · Markdown. Collect, draft, review, publish and record.
drafts.mjs
Script · Node. Maintains the private ledger and exports public summaries.
verify-publication.mjs
Script · Node. Checks the live article before adding its link.
build-in-public-runs.json
Registry · JSON. The public summaries read by this log.
Not packaged for public download yet. The guide on the left describes what each one does, step by step, so you can rebuild the setup; to have us run one, order it in the block below.
Run it for your company
The same agent, your buyers’ questions, your account.
Sign in, tell us your site in three lines, and we email you within one working day with how we would run it for you and what it costs. Nothing to install, nothing goes out without your approval.
Good to know
Where it stops and asks for you
- No useful change Records a skipped range without inventing an article.
- Drafting fails Records a failure and keeps the input range for retry.
- Website verification fails The log continues to show a draft awaiting publication.
Risk notice
What to watch for
- A commit proves a development change, not a successful deployment or a business result.
- Private drafts and diagnostics stay local. Back up the local ledger to preserve the processing position.
FAQ
Questions about this agent
Does saving a draft publish it?
No. The public log shows a draft awaiting review, without its text. Article publication is verified separately.
Will this send posts to LinkedIn, X or Medium?
No. Social distribution is on hold, and website articles opt out of automatic Medium syndication.
When does a run appear here?
The scripts update the public summary file locally. It appears on this site after that file is committed and deployed.
Why do the first two reports have the same run date?
Both were run on October 31. Each row names the development period it covers; the run date records when we did the work.
Would you rather we ran this agent for you? There is a managed service for that.
Talk to usOther agents