The Maintainer's continue or stop is recorded in a file, and a decision counts only in his name (#198's ruling 7, record 0194)
User description
This records the design's first decision point as a file, with what the Maintainer is shown before he decides. It is working ruling 7 on "The ruleset switch: main requires the machine user's approval of the most recent commit, and slice 1's event is recorded" (#198). Part of #198; it does not close it.
Model-Lens inputs touched: none. config/hooks.json is read, not changed.
Size. 48,096 characters and 725 lines by git diff; 52,705 characters as the Kernel counts a diff, with its lines numbered, against a budget of 60,000. Ruling 5's part, the skill's "When the Factory cannot approve", was split off to keep this under it and follows as its own pull request with a record of its own.
What changes
-
release/decision-points.json, of the new schemadecision-points/1: one entry,first_release_on_next. Its Release is factory/v0.1.6, the one that reviewed slice 1's event.decided,dayandbyare null: they are the Maintainer's to write.Shown to him Figure Pull requests merged into mainbefore the event merged, 2026-11-11T03:57:41Z387 Of them, his own 387 Of them, changed a protected path of config/hooks.json165 The design's starting figure 40 -
tools/check_release.pyholds what the schema cannot say:- a point is recorded once;
decided,dayandbyare written together or not at all;- a decision counts only where
byis the loginconfig/hooks.jsonnames the Maintainer by, in any case; where that file names nobody, nobody's counts; - the decision's day is the day the counts were read, or later;
- the counts are read on the event's day or after;
- each day written is a day of the calendar;
- no share is greater than the count it is of.
-
tools/check_examples.pyholds the file to its schema, and sends a counter-example the schema takes to that rule. -
Two counter-examples: a decision in another login's name, and a point with no count.
-
Record 0194 says the form, who decides, and what is not held.
What is not held, and the record says so
- Whose hand wrote a decision. The check holds the name written. Every commit names the Maintainer, so nothing in the repo tells his edit from a session's, and the file is not a path a person must approve. Four readers raised this. Making it one is a change to
config/hooks.json, a model Lens's input, so it is not in this pull request; it is put to the Factory Builder as its own small one. - The counts. They were read once and no check reads them again, since
verifyreads the commit alone:- GitHub's search, through the Maintainer's login:
is:pr is:merged base:main merged:<2026-11-11T03:57:41Zgives 387, and 387 withauthor:don-petry. - The repo's history: 390 commits on
main's first-parent line before the event's, 387 of them a pull request's squash and three the first commits. 165 of the 387 change a path theprotectedlist ofconfig/hooks.jsonnames, as it stood at 687db79b.
- GitHub's search, through the Maintainer's login:
What it does not do
- It does not decide.
decidedis null until the Maintainer writes it. - It adds no row to
release/AGENTS.md, which is at its 40-line limit. The record's boundaries carry the rules. - A count beyond the figure stops nothing, by the design.
Checks
| Check | Result |
|---|---|
make checks | OK |
tests.test_release, tests.test_evals | OK, 114 tests |
tests.test_tools, the tests of schemas, examples, envelopes, skills and decisions | OK, 28 tests, at the first head |
| Mutants of the seven touched files, run alone on an export | 102 of 102 killed, at 5b9d98e |
Full make verify | Not run locally, by the Maintainer's standing change of 2026-11-09; CI's verify is the answer |
| Reviewer | Head | Unread parts | Blocking | Major | Minor | Info |
|---|---|---|---|---|---|---|
| agy (Carbon) | 1bf54b3 | none of 8 | 0 | 7 | 6 | 0 |
| muse | 1bf54b3 | none | 1 | 1 | 0 | 0 |
| claude-review | 1bf54b3 | none | 0 | 0 | 3 | 1 |
| claude-security | 1bf54b3 | none | 0 | 1 | 0 | 0 |
| astra | 5f50124 | none | 0 | 2 | 0 | 0 |
The four first readers read 1bf54b3, which still held the skill's part. Astra read 5f50124, after their findings were fixed and the skill's part had left. The head adds Astra's two fixes, names four test tuples once in tools/mutants.py, and mends one finding of a review thread: two points that name none are no longer read as one recorded twice (5b9d98e). After that it merges main and renumbers the record from 0193 to 0194 (880c1b1), since "A run record says when its Run began, read from its jobs under P-64's rule (#724, record 0193)" (#751) merged first; nothing decided changes. By the lane's rule a gap in a test needs no read again, so no reader read again.
| Finding | What | Disposition |
|---|---|---|
| muse #1 | The record, then numbered 0193, left a gap: 0192 was not on main | fixed by the merge of #731, which main now holds |
| security #1, muse #2, claude-review #2, agy #1 | "Only the Maintainer writes it" is held as a string any session can type | fixed in the record: "Who decides" and the boundary say what the check holds, the name; "What is not held" says the hand is not. Protecting the path is put to the Factory Builder |
| claude-review #1, agy #8 | The hand approval is allowed only when no Release can serve, though a change the Factory may not approve also needs it | moved with the skill's part to the next pull request, which adds that case |
| claude-review #3, agy #5 | Type guards of the rule survive removal | fixed: each odd part alone, and a mutant for each guard |
| claude-review #4 | "A Pause stops the Factory's writes" | moved with the skill's part, and reworded there to the design's I-7 |
| agy #2 | The schema lets one of the three be written without the others | the check refuses it, as it did; the schema's title now says the check holds it |
| agy #9 | The counts' day is not held to the event's | fixed: a rule, its boundary and four mutants |
| agy #10 | Days are not said to be UTC | fixed in the schema and the record |
| agy #3 | No Maintainer login in another case is tried | fixed, with a mutant |
| agy #4 | Two points of two names are never shown to pass | fixed, with a mutant |
| agy #11 | A half-written decision never names another | fixed |
| agy #12 | The no-count example is not held to the live shares | fixed |
| agy #6, #7, #13 | The skill's test: the machine user, other code blocks, other steps | moved with the skill's part, and fixed there |
| astra #1 | A 31st of November passes as a day | fixed: each day is held to the calendar, with four mutants |
| astra #2 | Nothing takes a stop through the schema | fixed: both decisions through the schema and the check, with three mutants |
🤖 Generated with Claude Code
CodeAnt-AI Description
Record and validate the Maintainer’s first release decision
What Changed
- Records the first release decision point with the context shown to the Maintainer: 387 merged pull requests, all 387 authored by him, 165 touching protected paths, compared with the design’s starting figure of 40. The decision fields remain empty until he records a choice.
- Accepts a decision only when all decision fields are completed together and the recorded login matches the Maintainer named in the hooks configuration.
- Rejects duplicate decision points, invalid dates, counts read before the event, decisions made before the counts were read, and shares larger than the total.
Impact
✅ First release decision has its counts and context in one record
✅ Partial or non-Maintainer decisions are rejected
✅ Decision dates cannot predate the counts shown
💡 Usage Guide
Checking Your Pull Request
Every time you make a pull request, our system automatically looks through it. We check for security issues, mistakes in how you're setting up your infrastructure, and common code problems. We do this to make sure your changes are solid and won't cause any trouble later.
Talking to CodeAnt AI
Got a question or need a hand with something in your pull request? You can easily get in touch with CodeAnt AI right here. Just type the following in a comment on your pull request, and replace "Your question here" with whatever you want to ask:
@codeant-ai ask: Your question here
This lets you have a chat with CodeAnt AI about your pull request, making it easier to understand and improve your code.
Example
@codeant-ai ask: Can you suggest a safer alternative to storing this secret?
Preserve Org Learnings with CodeAnt
You can record team preferences so CodeAnt AI applies them in future reviews. Reply directly to the specific CodeAnt AI suggestion (in the same thread) and replace "Your feedback here" with your input:
@codeant-ai: Your feedback here
This helps CodeAnt AI learn and adapt to your team's coding style and standards.
Example
@codeant-ai: Do not flag unused imports.
Retrigger review
Ask CodeAnt AI to review the PR again, by typing:
@codeant-ai: review
Check Your Repository Health
To analyze the health of your code repository, visit our dashboard at https://app.codeant.ai. This tool helps you identify potential issues and areas for improvement in your codebase, ensuring your repository maintains high standards of code health.