Make the contributor a standing member of every project's audience

This commit is contained in:
2026-10-21 22:23:55 +02:00
parent 6f58a0e5c2
commit 2407689747
5 changed files with 11 additions and 6 deletions
+1
View File
@@ -2,6 +2,7 @@
## Unreleased
- Every project's audience now includes the contributor, the person who reads the code to change it, and larv-review's comprehension check names them as the reader it serves. A README doesn't list them; only its `**Void:** code health` line drops them.
- `settings.json` no longer carries a model: each machine keeps the one you pick with `/model`. Git's clean filter `bin/settings_filter.py` drops `"model"` and sorts the keys, so neither a pick nor Claude Code's key order can reach a commit. Update now starts with `git add settings.json`: without it, Update's `git checkout -f` drops the machine's model. On upgrade, on each machine:
1. Turn the filter on in `~/.claude`: `git config filter.settings.clean bin/settings_filter.py`.
2. Pick a model with `/model`. Until you do, Claude Code runs its default model; this repo set `opus` before.
+8
View File
@@ -697,3 +697,11 @@ Read most (opens and searches):
## Reading cost, chunk dd5ad8ab4fdf, finished 2026-10-21T18:27:48Z
58208 new input tokens before the first commit.
## Reading cost, chunk 14bbaa6d288c, finished 2026-10-21T20:23:13Z
78001 new input tokens before the first commit.
Read most (opens and searches):
- `rules/decisions.md` 1
+1
View File
@@ -9,3 +9,4 @@ paths:
- **Mark each goal `**Maintain:**` or `**Achieve:**`.** A maintain goal is a state to hold: a change that breaks it is redesigned or dropped, whatever its value to other goals. An achieve goal is a direction to pursue, weighed against the other achieve goals. Under each achieve goal, three indented bullets: `Signal:` what moving towards it looks like, `Counter-signal:` what shows it is bought at a wrong price, and `Why:` one line on how the two connect. A goal with sub-goals states its own signal. A maintain goal has no bullets. BANNED: an unmarked goal; an achieve goal with no signal; `**Achieve:** Delivery speed.` with the signal `faster`. GOOD: `**Achieve:** Delivery speed.` with `Signal: hours from a chunk's start to its merge fall.`, `Counter-signal: 🔴 findings per chunk rise.` and `Why: speed bought by skipping review shows up as defects.`
- **Leave code health out of the README.** `~/.claude/CLAUDE.md` → "Shared goal" states it for every project, ranked first, and says when and how a README voids it. File an item to drop a README goal that restates it: the goals after it renumber, and every Moves cell and decision citing them must follow.
- **Audience** names who depends on the project and what they may rely on.
- **The contributor is in every project's audience**, whoever the README names: the person who reads the code to change it. Don't list them. A README drops them only with the `**Void:** code health` line (`~/.claude/CLAUDE.md` → "Shared goal"), the one place a project says maintainability doesn't matter.
+1 -1
View File
@@ -114,7 +114,7 @@ Launch principles-auditor on the trigger in §3: one for the README's goals and
### Comprehension check
The comprehension check records how hard the project is to read, and blocks no merge. Simulated readers, the seats, launched per `~/.claude/skills/comprehension-panel/SKILL.md`, each read the whole project, not the diff. It reports no findings and is no pass: it runs once, in §5, after every pass is done, the auto-reviewer included, so it reads the branch as review left it. It runs a paired ruling when the architecture pass's trigger fired, whether or not a precedent, the user's no or an opt-out skipped its architect, and a scoring run where one is due. It skips only as "Opt-out" or "Suspended" below says.
The comprehension check records how hard the project is to read for the contributor, the reader it serves (`~/.claude/rules/goals-and-audience.md`), and blocks no merge. Simulated readers, the seats, launched per `~/.claude/skills/comprehension-panel/SKILL.md`, each read the whole project, not the diff. It reports no findings and is no pass: it runs once, in §5, after every pass is done, the auto-reviewer included, so it reads the branch as review left it. It runs a paired ruling when the architecture pass's trigger fired, whether or not a precedent, the user's no or an opt-out skipped its architect, and a scoring run where one is due. It skips only as "Opt-out" or "Suspended" below says.
**Paired ruling.** Each seat votes whether the project reads better, the same or worse at this branch's head than at the base.
-5
View File
@@ -12,7 +12,6 @@
| ID | Release | Blocked by | Exempt | Item | L | E | I | R | S | Moves | Score |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 29 | 0.3.0 | | | **Make the contributor a standing member of every project's audience, unless its README voids code health.** | 2 | 2 | 3 | 5 | 2 | 1.1 by 5, 1.4 by 5 | 13.7 |
| 199 | 0.3.0 | | defect | **Start loop-claude's worker with `ScheduleWakeup` disallowed, so a worker that ends its turn on a question exits and the judge runs.** | 2 | 1 | 3 | 4 | 2 | 1.5 by 6, 1.2 by 3 | 11.8 |
| 170 | 0.3.0 | | decision | **Re-tune `WORKER_STEPS` and `REVIEW_STEPS` once 10 chunks have shipped under them, from how well L, E and I order their review rounds and 🔴 findings.** | 2 | 2 | 3 | 5 | 2 | 1.1 by 3, 1.4 by 5 | 10.9 |
| 117 | 0.4.0 | | | **Give `todo_table.items` one way to say a plan has no items table, so no caller tells it from an empty plan by truthiness.** | 1 | 1 | 3 | 3 | 2 | code health by 4, 1.4 by 4 | 29.2 |
@@ -91,10 +90,6 @@
## Details
### 29. Make the contributor a standing member of every project's audience, unless its README voids code health.
State that in `~/.claude/rules/goals-and-audience.md` → "Goals and Audience", and have larv-review's comprehension check name the contributor as the reader it serves. The `**Void:** code health` line (`~/.claude/CLAUDE.md` → "Shared goal") is the one place a project says maintainability does not matter: it drops the contributor. A project that opts out of the comprehension check must already write that line.
### 199. Start loop-claude's worker with `ScheduleWakeup` disallowed, so a worker that ends its turn on a question exits and the judge runs.
Claude Code's `ScheduleWakeup` tool resumes a session after a delay. While a wakeup is pending, the worker's `claude -p` keeps running after its turn ends. loop-claude acts only once the worker exits, so it neither runs the judge nor interrupts the human, and the chunk stalls. Seen on 2026-10-18 in another project, session `678cac80`: the worker asked the human to approve a fourth round of the stability pass, then scheduled a 30-minute wakeup, and another each time one fired, until someone killed the worker. A worker doesn't need a wakeup: larv-review already has it wait on reviewers and CI in the foreground. Add `--disallowedTools ScheduleWakeup` to `bin/loop_claude/cycle/worker.py`'s `worker_args`, with a test asserting the flag. Correct `skills/larv-review/SKILL.md`'s opening paragraph, which says a scheduled wakeup dies with a headless run.