Standing service outside-contributor-drafts-sampled-and-retested-every-month · revised 12 October 2026
Standing service
Have a sample of your contributor-reply drafts read and re-tested every month
Each month a sample of the drafts is read against your guide, the hostile-contribution set is re-run on a copy, and you get a report. Nothing is posted or changed for you.
This starts a conversation by email. Nothing is charged, and nothing is monitored, until we have agreed scope, access and terms with you in writing.
The responsibility you hand over
A reply-drafting job that matched your guide at launch drifts as the guide, the templates and the model change. Maintainers stop reading drafts closely, and the first wrong promise is posted under the project's name.
Who it’s for: Maintainer whose reply-drafting job is live and who wants proof it still drafts safely.
Usually starts when: Your guide changed, a maintainer posted a draft that needed heavy edits, or a month passed with nobody checking.
The result: Every month a random sample of 30 drafts is rated against your guide, the promise and legal-phrase checks are re-run over all drafts, the hostile set is re-run on a copy, and you receive a report with what to change in the guide or the rules.
What stays true, and what we do about it
No response-time guarantee is published for this new service. A date for each monthly report is agreed in writing before it starts.
Hours are agreed in writing before the service starts. It runs once a month and is not staffed round the clock.
What must remain true
- The report flags any sampled draft promising a merge, delivery deadline or feature, and reports all script flags for licence or legal phrases
- The report records whether each sampled draft has a valid guide citation or says no source was found
- The report includes every hostile-set result and flags any marker, forbidden action or changed result
What we watch
- A monthly export of drafts and the posted replies
- The contributing guide and templates
- Your list of rule and guide changes
When something happens
Scroll the table sideways to read it all.
| When | What we do |
|---|---|
| A month closes and the export is available. | We rate the sample, check every draft by script, re-run the hostile set and send the report. |
| You change the guide, a template or the draft rules. | We re-run the replay and hostile sets on a copy before you rely on the change. |
| A maintainer edits most of a draft before posting. | We list the pattern in the report and suggest a guide or rule change. |
We do on our own
- Rate the sampled drafts
- Check every draft by script
- Re-run the hostile set on a copy
- Write the report
We ask you first
- Any request to read private messages or security reports
- Any change to a rule, setting or page in your own systems
- Any wider access than the exports named here
- Contacting your customers or anyone outside your team
We escalate to you when
- A draft that promises a merge, a date or a feature
- A hostile contribution that changed a draft
- A draft was created for a security report
- The off switch fails on the copy
How you know it held. Your maintainer holds marks for 10 sampled drafts until our report is frozen; at least 9 agree, and any unflagged forbidden promise, legal claim or security-report draft fails.
How we keep it true
This service is never finished. Each month's report shows whether the things it watches stayed true, and it continues until you end it.
Get draft replies to outside contributors' issues and PRs, for a maintainer to post Job Optional
New issues and pull requests from outside contributors get a draft reply that checks them against your contributing guide. A maintainer reads and posts each one. The job never posts, labels or closes.
Bought separately, each time it is needed; not part of the monthly fee
The one-off job; buy it first if you do not yet have one.
What is included, and what is not
- The monthly report
- The sample ratings
- The phrase-check results
- The hostile-set re-run
Included
- One repository and the job built under the one-off job on this site or one you can show us
- A random sample of 30 drafts, rated for guide fit, with the seed recorded
- A script check of every draft in the month for promises, dates and legal phrases
- A re-run of the hostile set on a copy after any change you report
- A monthly report with counts and guide changes to consider
Not included
- Posting, labelling or closing anything
- Holding your token
- Deciding what to accept
- Replies to security reports
- Any promise about response times
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Your maintainer selects 10 of the 30 sampled drafts and keeps their marks from us until the report is frozen; at least 9 marks agree under the pre-agreed guide-fit rubric. One unflagged promised merge, delivery deadline, feature, legal claim or draft for a security report fails the whole test; this is a reporting smoke test, not a future rate.
Evidence: The seed, the 10 drafts and the maintainer's marks beside ours.
The report states the sample size, the number of drafts the script flagged and what it cannot show.
Evidence: The monthly report.
The hostile set is re-run on a copy each month and any change in results is listed first.
Evidence: The re-run results.
Nothing has been posted, labelled or changed on your repository by us.
Evidence: Your repository's audit log for the month.
Sign-off. You read each report and re-run the acceptance checks with the files and steps supplied. A month counts as delivered only when those checks pass.
If it fails. If we cannot do a month because an export is missing or has changed, we say so and do not guess. If the service is not working for you, you can end it at the end of any month.
When it fits, and when we stop
It fits when
- A reply-drafting job with a queue export containing at least 30 drafts in the month
- The current contributing guide
- A maintainer who rates the sample
We stop and tell you if
- The queue export is missing
- The guide changes and the rules are not updated for two months
- No maintainer can rate the sample
What could go wrong
The service reads exports and reports, so there is nothing to undo. Ending it leaves your systems exactly as they are; you keep every report.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A sample of 30 misses a rare error. | The report states the sample size and that a sample is a tripwire. The script still checks every draft. |
| Exports carry contributors' personal details. | Source contributions and posted replies are public, but unposted drafts are private queue data. You remove unnecessary personal details before exporting; written terms cover any model service and deletion of our copies. |
Each report is checked by a separate automated reviewer before hand-over. A maintainer on your side re-rates a sample of drafts and decides every change. No human supervisor is included unless your agreement names one, and at launch the work is largely automated.
Stays with a person
- Your maintainers change every rule, guide section and setting
- You hold the token and the off switch
Access we would need
- A monthly export of the draft queue and the posted replies
- A copy of the job for the re-runs
Questions
Do you post or close anything?
No. We read exports and report. Your maintainers do everything on the repository.
Do you promise faster replies?
No. We never state response-time or contributor-count figures.
Send an enquiry
Send us
- Which repository and job
- Roughly how many outside contributions arrive a month
- No logins, keys, customer data or payment details in the first enquiry
Later, once you agree
- A monthly export of the draft queue with the replies actually posted
- The current guide and templates
- Your list of guide or rule changes during the month
You keep the repository and job; we receive the agreed queue and reply exports, work on a copy, never hold your token and never post or change repository content. Before sharing, the terms name any model service, its country, retention and training status, and when our copies are deleted.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “outside-contributor-drafts-sampled-and-retested-every-month” as the subject.