Record the comprehension check: two scoring runs at 8f03747, overall 6.25 and 6.00; refresh floor item 300
CI / tests (pull_request) Failing after 2m20s

This commit is contained in:
2026-11-13 02:24:47 +01:00
parent 8f03747d9a
commit 6b7392adcb
2 changed files with 125 additions and 27 deletions
+103
View File
@@ -3686,3 +3686,106 @@ Ruling: same
|---|---|---|---|---|---|
| Mid A | same | same | same | same | same |
| Inherited architect | same | same | same | same | same |
## 2026-11-13T01:19:17Z, PR #520 at 8f03747
| Seat | Navigation | Locality | Shape | Self-sufficiency | Overall |
|---|---|---|---|---|---|
| Junior A | 7 | 5.5 | 6 | 5.5 | 6 |
| Mid A | 6.5 | 5.5 | 6 | 5 | 6 |
| Maintainability senior | 7 | 6 | 6 | 6 | 6.5 |
| Inherited architect | 7 | 6 | 6 | 6.5 | 6.5 |
| Mean | 6.88 | 5.75 | 6.00 | 5.75 | 6.25 |
Junior A, hardest first:
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`: five notions of rung plus the escalation mark and `UNFINISHED_PER_STEP`, with each step's reason in the decision log
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: stamps, annotates, steps up and archives in an order only the code states, with `close.py` archiving again
- Locality, Shape: `bin/release_plan/ranking.py:304` `score`: every function forks on `SHAPE` versus `OLD_SHAPE`, with no word on when the old one goes
- Shape: `bin/release_plan/release_standing.py:168` `next_release`: six `Pick` types and `| Failure` or `| None` fields in `LastCut`
- Shape: `bin/rule_hook/todo_check.py:22` `todo_breaches`: three config paths, and why the checkout's own script runs only under `same_repo` stays unclear
- Shape: `bin/loop_claude/loop/reload.py:220` `decoded_route`: one dense guard, and renaming a `LoopRun` field stops a running loop
- Self-sufficiency: `skills/larv-review/SKILL.md:82` "Round cap": needs `CLAUDE.md`, `review-findings.md` and the rules files open at once
Mid A, hardest first:
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`: five rung notions across four helpers and the escalation mark, the cap's reason only in the decision log
- Shape, Locality: `bin/rule_hook/todo_check.py:22` `todo_breaches`: four paths, `held()` and `located()` answering similar questions differently, and `changed` hashing a blob by hand
- Shape: `bin/release_plan/release_standing.py:168` `next_release`: three-way `LastCut` fields whose `None` means different things per field
- Locality: `bin/release_plan/ranking.py:304` `score`: two table shapes branch through scoring, and the weights' reasons sit elsewhere
- Locality: `bin/loop_claude/cycle/judge.py:196` `quotes_checked`: six checks each turn the verdict into something different
- Shape: `bin/larv_review/plan.py:66` `plan_reasons`: several failing reads threaded through every branch, and `ids=None` means something only the callers say
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: a fixed order nothing enforces, and an 18-field `Cycle` passed everywhere
- Self-sufficiency: `bin/larv_review/main.py:1` help: about 150 lines of rules a reader needs before any handler
Maintainability senior, hardest first:
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`: six things held at once, and why the rules combine sits behind the decision log
- Shape: `bin/rule_hook/todo_check.py:42` `folder_breaches`: four cases and a second fork in `check_failures`, and `held` neither says what it tells nor that `config.top` is the main checkout
- Shape: `bin/rule_hook/rules.py:82` `todo-rank-check`: one help sentence carries three conditions joined by "and"
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: archives twice in a set order, and `stepped_to` changes `cycle.route` outside the state machine
- Locality, Self-sufficiency: `bin/release_plan/ranking.py:160` `weigh`: two table formats, and constants whose reason is only "docs/decisions.md records the choice"
- Shape: `bin/rule_hook/main.py:116` `bash_breaches`: a 200-character return line joining four kinds of breach
- Locality: `bin/larv_review/main.py:1` help: the only full statement of `needs-scoring-run` and `scoring-run-owed`, implemented in other modules
Inherited architect, would restructure first:
- Locality, Shape: `bin/chunk_preflight/choices.py:80` `forge_of`: forge code split across `chunk_preflight` and `deploy_release`, which three tools reach into
- Navigation: `bin/records/__init__.py:1` `records`: file utilities share a name with the judge's records in `loop_claude/project/records.py`
- Shape: `README.md:3` dated experiments: five research and outreach folders at the root beside the working areas
- Navigation, Shape: `README.md:131` `bin/`: 66 entries at one level, navigable only through the README's prose map
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: `rule`, `close` and `keep_rejection` each archive and map a Failure to `Stop(1)`
- Self-sufficiency: `bin/rule_hook/round_cap.py:26` `earlier_titles`: fails open when git cannot list the branch's titles, so a mislabelled round passes
## 2026-11-13T01:23:02Z, PR #520 rerun at 8f03747
| Seat | Navigation | Locality | Shape | Self-sufficiency | Overall |
|---|---|---|---|---|---|
| Junior A | 6.5 | 5 | 5.5 | 5 | 5.5 |
| Mid A | 7 | 5.5 | 6 | 5.5 | 6 |
| Maintainability senior | 7 | 6 | 6 | 6 | 6.5 |
| Inherited architect | 6.5 | 6 | 5.5 | 6 | 6 |
| Mean | 6.75 | 5.63 | 5.75 | 5.63 | 6.00 |
Junior A, hardest first:
- Shape: `bin/release_plan/release_standing.py:168` `next_release`: a before-first-release constant, cut and named releases, and `str | None | Failure` fields tracked to where each is checked
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`: the escalation rule split across four files, five ideas in a few lines, and the reasons in the decision log
- Shape: `bin/rule_hook/todo_check.py:22` `todo_breaches`: three paths, the checkout's own script only under `same_repo`, and a blob hashed by hand
- Locality: `bin/loop_claude/cycle/run.py:67` `run_cycle`: changes `cycle.route` away from where the 17-field `Cycle` is built
- Locality: `bin/larv_review/owed.py:46` `shortfall`: logic hangs on the exact wording of a sentence SKILL.md also depends on
- Shape: `bin/rule_hook/committed_files.py:14` `committed_file_breaches`: a reversed-dict dedupe whose choice of copy no comment explains
- Self-sufficiency: `skills/larv-review/SKILL.md:82` "Round cap": every sentence points to another rule, and the code read easier than the prose
- Shape: `bin/release_plan/ranking.py:315` `weight_of`: old and new table shapes in five functions, and a docstring packing three rules into one sentence
Mid A, hardest first:
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`: five things held for one step-up, and why the rules interact lives in the decision log
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: archives twice in an order that matters, repeated across `run.py` and `close.py`, and nothing guards it
- Self-sufficiency: `bin/release_plan/release_standing.py:168` `next_release`: `BEFORE_FIRST_RELEASE` and the earliest test need CLAUDE.md's "Working loop" open beside them
- Shape: `bin/release_plan/ranking.py:56` `rank`: each function reads as two algorithms interleaved, old shape and new
- Self-sufficiency: `bin/larv_review/plan.py:66` `plan_reasons`: the 155-line help and the code each state part of when a scoring run is due
- Self-sufficiency: `bin/loop_claude/cycle/judge.py:21` `INTERRUPT_RULES`: behaviour held in prompt text and schema descriptions the code can't explain
- Shape: `bin/rule_hook/todo_check.py:22` `todo_breaches`: three branches, and a subprocess path I would not have guessed
- Self-sufficiency: `skills/larv-review/SKILL.md:82` "Round cap": dense rules pointing at each other, some at files out of reach
Maintainability senior, hardest first:
- Shape, Self-sufficiency: `bin/release_plan/release_standing.py:168` `next_release`: `Failure` in every `LastCut` field, a sentinel version and a `nothing_cut` flag worked through as truth tables
- Locality: `bin/loop_claude/routing.py:93` `unfinished_route`: five helpers and the escalation mark held at once, the rule kept in no one place
- Shape: `bin/release_plan/ranking.py:160` `weigh`: the legacy shape doubles every scoring function, and nothing says when it can go
- Shape: `bin/rule_hook/committed_files.py:14` `committed_file_breaches`: a double-reverse dedupe with no comment
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`: five steps in a row, overlapping archive helpers, and an order that is the trap
- Shape: `bin/rule_hook/todo_check.py:42` `folder_breaches`: `config.folder` to `changed` but `config.top` to `check_failures`, right but a stop to confirm
- Shape: `bin/model_routing/route.py:28` `levels_fit`: two rules in one boolean, and `route_of` packs its type checks into one long line
- Self-sufficiency: `skills/larv-review/SKILL.md:82` "Round cap": dense bullets pointing elsewhere, reread before the code matched
Inherited architect, would restructure first:
- Navigation, Shape: `README.md:131` `bin/`: about 63 entries at one level, with the shared libraries told apart only by README prose
- Locality, Shape: `bin/chunk_preflight/choices.py:80` `forge_of`: forge identity and forge calls in two tool packages, so a second forge lands in three
- Navigation: `bin/records/__init__.py:1` `records`: a utility package named like the judge's records, and quote semantics in `claude_code/transcript.py:85` `verbatim`
- Locality: `bin/loop_claude/cycle/run.py:94` `archived`: archive helpers in the cycle runner, which `loop/close.py` imports
- Shape: `README.md:3` trial folders: four finished experiments at the root beside the working areas
- Locality: `bin/rule_hook/rules.py:61` `round-cap`: the cap's wording spelled in SKILL.md, `rules.py` and `round_titles.py`
- Self-sufficiency: `bin/rule_hook/round_cap.py:26` `earlier_titles`: fails open on the unattended path when git cannot list the branch's titles
+22 -27
View File
@@ -21,7 +21,7 @@
| 349 | 0.11.0 | | | **Convert markdown to ADF and back with `@larvit/adf-codec`'s portable calls, through a pinned Node shim run in a container with a memory limit.** | 2 | 1 | 2 | 3 | 2 | 1.6 by 3 | 1.1 |
| 351 | 0.11.0 | 348, 349, 350 | | **Read a project's plan from the child issues of a Jira Cloud epic its config names, ranked in Jira's backlog order.** | 3 | 2 | 3 | 6 | 5 | 1.6 by 6 | 1.2 |
| 353 | 0.11.0 | 351 | | **Let a project's config take its goals and audience from a Jira epic's description, under the same two headings.** | 2 | 2 | 3 | 5 | 3 | 1.6 by 4 | 1.3 |
| 358 | 0.11.0 | | | **Let a project's config allow its runtime and tests on the host, and set the comprehension floor that marks code health very important.** | 2 | 2 | 3 | 5 | 3 | 1.6 by 4 | 1.3 |
| 358 | 0.11.0 | | | **Let a project's config allow its runtime and tests on the host, and set the comprehension score under which code health is very important.** | 2 | 2 | 3 | 5 | 3 | 1.6 by 4 | 1.3 |
| 362 | 0.11.0 | 348 | | **Send loop-claude's notices through the notifier a project's config names, with Mattermost as the first backend.** | 2 | 1 | 2 | 3 | 3 | 1.6 by 3 | 1.0 |
| 363 | 0.11.0 | 362 | | **Add a Slack backend to the notifier, sending each notice as a DM from a bot.** | 2 | 1 | 2 | 3 | 3 | 1.6 by 4 | 1.3 |
| 352 | 0.11.0 | 351 | | **Write a project's plan to its Jira epic: file items as child issues, edit those the loop filed, score unscored issues with a comment, and move a finished item to the status its config names.** | 3 | 2 | 3 | 6 | 5 | 1.6 by 6 | 1.2 |
@@ -56,7 +56,7 @@
| 273 | 0.14.0 | | decision | **Re-tune the E + I steps of `REVIEW_STEPS` for 6 to 8 once 10 finished chunks have scored an E + I of 6 or more, from how well E + I orders their review rounds and 🔴 findings.** | 1 | 2 | 2 | 3 | 2 | 1.1 by 2, 1.4 by 3 | 2.0 |
| 268 | 0.14.0 | | | **End a run whose release is ready to cut with a notice on loop-claude's terminal, in place of an interrupt that waits for an answer.** | 2 | 2 | 3 | 5 | 3 | 1 by 5 | 1.9 |
| 171 | 0.14.0 | | decision | **Group the routing figures' review rounds by the reviewers' rung as well as the worker's, so a call blames a rung only for rounds its own reviews raised.** | 3 | 2 | 2 | 5 | 4 | 1.2 by 3, 1.4 by 3 | 1.8 |
| 300 | 0.14.0 | | owner | **Reach the comprehension floor: restructure the places the comprehension seats found hardest to follow, so the project reads at 7.0 or above on every dimension, Self-sufficiency first; today the scoring run at fba48f3 scored Navigation 6.88, Locality 5.63, Shape 6.00, Self-sufficiency 5.00 and Overall 5.88.** | 4 | 2 | 3 | 7 | 8 | code health by 8 | 1.7 |
| 300 | 0.14.0 | | owner | **Reach the comprehension floor: restructure the places the comprehension seats found hardest to follow, so the project reads at 7.0 or above on every dimension, Locality and Self-sufficiency first; today the scoring run at 8f03747 scored Navigation 6.75, Locality 5.63, Shape 5.75, Self-sufficiency 5.63 and Overall 6.00.** | 4 | 2 | 3 | 7 | 8 | code health by 8 | 1.7 |
| 264 | 0.14.0 | | | **Report the comprehension seats and the escalated subagents as subagent types of their own in `routing-figures.md`, with their dollars.** | 2 | 2 | 1 | 3 | 3 | 1.4 by 5 | 1.7 |
| 323 | 0.14.0 | | | **Count a release review as landed once its pull request has merged, whatever its commits are titled.** | 3 | 2 | 3 | 6 | 3 | 1.4 by 3, 1.5 by 2 | 1.7 |
| 342 | 0.14.0 | | | **Research the tooling agents read, search and run commands with once more, and file an item for each tool that cuts tokens a replay shows keep quality.** | 1 | 2 | 1 | 2 | 3 | 1.4 by 3, 1.5 by 2 | 1.7 |
@@ -123,7 +123,7 @@ lilleman's calls, 2026-11-12. One epic is one project's plan. An item is a child
lilleman's call, 2026-11-12: this client keeps its goals and audience in the plan epic's description. The source is a config setting, never hard-coded; item 346 builds the file source this one adds to. Read through the adf-codec shim.
### 358. Let a project's config allow its runtime and tests on the host, and set the comprehension floor that marks code health very important.
### 358. Let a project's config allow its runtime and tests on the host, and set the comprehension score under which code health is very important.
lilleman's calls, 2026-11-12. Legacy products may have no container setup, so the config may allow host tooling; `CLAUDE.md`'s container rule and rule-hook then follow it. In 0.10.0 a decision in the project's records can put tooling on the host instead. The comprehension panel will score legacy code low, and today code health under 7.0 is very important. The config sets that floor, 7.0 in the configs of the projects under `~/code`; the signal stays that the score rises.
@@ -347,36 +347,31 @@ Releases are the human's, and `CLAUDE.md` → "Working loop" already ends the lo
Raised by the goals audit of PR #296. `bin/routing_figures/calls.py` groups a chunk's review rounds by the rung its worker started on, and calls for a routing change when a lower rung took more of them. Since PR #296 the reviewers may run above the worker's rung, on the one their item's E + I picks, and no figure groups rounds by it, though the archived worker line names the E + I it follows from. A cheap worker on an item of high E + I gets stronger reviewers, who may raise more rounds, and a call could then blame the worker's rung for them: an interrupt of lilleman that settles nothing (goal 1.2).
### 300. Reach the comprehension floor: restructure the places the comprehension seats found hardest to follow, so the project reads at 7.0 or above on every dimension, Self-sufficiency first; today the scoring run at fba48f3 scored Navigation 6.88, Locality 5.63, Shape 6.00, Self-sufficiency 5.00 and Overall 5.88.
### 300. Reach the comprehension floor: restructure the places the comprehension seats found hardest to follow, so the project reads at 7.0 or above on every dimension, Locality and Self-sufficiency first; today the scoring run at 8f03747 scored Navigation 6.75, Locality 5.63, Shape 5.75, Self-sufficiency 5.63 and Overall 6.00.
Filed by the comprehension check of PR #473 at 13d1550 (`larv-review record-scoring`), which printed `Floor:` and `Next:`, and refreshed from the second run of PR #510 at fba48f3. The lowest mean is Self-sufficiency, 5.00: all four seats name a reason that sits in `docs/decisions.md`, in the vocabulary at the end of `CLAUDE.md` or in `skills/larv-review/SKILL.md` rather than beside the code. Refresh this item's places and aim from the newest run before the chunk starts; do not file a second floor item.
Filed by the comprehension check of PR #473 at 13d1550 (`larv-review record-scoring`), which printed `Floor:` and `Next:`, and refreshed from the second run of PR #520 at 8f03747. The lowest means are Locality and Self-sufficiency, 5.63 each: the seats name state shared across modules, such as the route and the archive order, and reasons that sit in `docs/decisions.md`, `CLAUDE.md` or `skills/larv-review/SKILL.md` rather than beside the code. Refresh this item's places and aim from the newest run before the chunk starts; do not file a second floor item.
Aim: one ambitious refactor of the areas below, sized so every mean reaches 7.0 on the chunk's own scoring run. Where a place is already easier to read when the chunk starts, drop it.
Places the seats named, hardest first (`larv-review places`):
- Locality, Shape: `bin/loop_claude/routing.py:93` `unfinished_route`: the floor rung, the escalation cap, the escalation mark, the unfinished count and the taken-in E + I are held at once, across about six helpers in five files. The mark's meaning lives in `escalation_mark.py`, and the reason for the floor and the cap sits in the decision log.
- Locality, Shape: `bin/release_plan/release_standing.py:145` `next_release`: the tag, changelog, before-first-release and renamed branches meet in the `is_earliest` expression, which took a senior three reads. Nothing comments the `None` and `nothing_cut` case.
- Locality: `bin/loop_claude/cycle/judge.py:211` `quotes_checked`: the judge's contract is split over prose constants, schema descriptions and checks after the fact. A bad quote silently changes the verdict's type.
- Locality, Shape: `bin/loop_claude/cycle/run.py:146` `rule`, with `loop/close.py:18` `closed_cycle`: the order of the ruling, rejection and step-up archive writes carries meaning, and nothing enforces it. Four near-identical archive helpers, `run.py:86` `archived` among them, sit in the cycle runner though `loop/close.py` imports them, and `run_cycle` replaces the route partway through its loop.
- Locality, Self-sufficiency: `bin/larv_review/reviewer_header.py:49` `unanswered_round` and `skills/larv-review/SKILL.md:76` "Round cap": the cap counts answer commits, so a worker that relaunches without committing gets the same round number. The cap is enforced in four units, split over SKILL.md and three modules.
- Locality, Shape: `bin/larv_review/plan.py:66` `plan_reasons`: a merge-base commit enters the parameter that item 296's chunk renamed `ref`, eight failure branches mix git reads, file reads and logic, and `after` means the checkout's text and then its items.
- Locality: `bin/larv_review/owed.py:37` `shortfall`: the one-run-or-two rule is keyed on the wording `EMPTIED_WORDS` that plan.py builds and SKILL.md quotes, and `scoring_runs` defines a scoring run differently from `Entry.scored`. Give the scoring-run logic one module, and have the number of runs owed come from a value, not from a sentence.
- Locality: `bin/chunk_pr/main.py:42` `forge`: the forge helpers live in `chunk_preflight` and `deploy_release`, so a second forge needs edits in four packages.
- Locality: `bin/rule_hook/reviewer_rung.py:40` `skill_breaches`: three cases rest on assumptions made in another function.
- Locality: `bin/rule_hook/repeat_read.py:119` `shown_lines`: it rebuilds from transcript entries which lines are in the agent's context.
- Locality: `bin/model_routing/route.py:88` `route_of`: a route has two serializations, a flat archive line and a nested handover.
- Self-sufficiency, Locality: `bin/loop_claude/cycle/prompts.py:59` `review_launches` and `:119` `escalation`: the worker's prompt is built from about 20 conditional sentence builders, which restate the routing logic.
- Shape: `bin/release_plan/ranking.py:56` `rank`, `:304` `score` and `:315` `weight_of`: nearly every function forks on `SHAPE` versus `OLD_SHAPE`, which doubles the module; item 156 removes the old shape. The constants read as magic until `rules/todo-items.md` is open.
- Shape: `bin/review_rules/comprehension_history.py:127` `entries`: a line-by-line parser that tolerates two legacy formats.
- Shape: `bin/records/__init__.py:1` `records`: one name for three concepts, as `cut` and `Plan` also name several.
- Shape: `bin/rule_hook/rules.py:25` `RULES`: rules sit inline in one module while five others have a module each.
- Navigation, Shape: `README.md:116` `bin/`: about 64 entries in one flat list, some 35 of them packages, the worst fan-out level.
- Self-sufficiency: `skills/larv-review/SKILL.md:147` "Scoring run": dense rule prose with many cross-references; three sections rebuild one rule.
- Locality: `bin/review_rules/round_titles.py:1` `ROUND`: the commit-title protocol `Answer <pass> round <n>: …` is parsed in `round_titles.py`, `rule_hook/round_cap.py`, `score_outcomes`, `branch_chunks` and `routing_figures`, and written in SKILL.md.
- Locality: `bin/larv_review/scoring_setup.py:156` `prepare_scoring`: code parses the headings and fences of SKILL.md and `seat-prompts.md`, so a prose edit can break it.
- Shape: `bin/command_lines/shell_command.py:46` `segments`: a hand-written shell tokenizer whose regexes stay opaque.
- Shape: `reviewer-merge-trials/driver.py:1` `driver`: one-off trial scripts space-indented and tersely named, against the repo's tabs and naming.
- Locality, Self-sufficiency: `bin/loop_claude/routing.py:93` `unfinished_route`, with `escalated_route`: the floor, the cap counted from `item_rung`, the escalation mark, the unfinished count and the E + I taken in are held at once across four files, and why they combine sits in the decision log. Every seat named it; three would least want to change it.
- Shape, Self-sufficiency: `bin/release_plan/release_standing.py:168` `next_release`, with `standing`: `str | None | Failure` fields in `LastCut`, the `BEFORE_FIRST_RELEASE` sentinel and `nothing_cut` meet in the `is_earliest` test, read as truth tables; four tools choose the next release through it.
- Locality: `bin/loop_claude/cycle/run.py:154` `rule`, with `:67` `run_cycle` and `:94` `archived`: archives twice in an order that matters and that nothing enforces, the archive helpers sit in the cycle runner though `loop/close.py` imports them, and `run_cycle` replaces `cycle.route` outside the state machine.
- Shape: `bin/release_plan/ranking.py:56` `rank`, `:160` `weigh`, `:304` `score` and `:315` `weight_of`: every function forks on `SHAPE` versus `OLD_SHAPE`, and the constants' reasons are only "docs/decisions.md records the choice".
- Shape: `bin/rule_hook/todo_check.py:22` `todo_breaches`, with `:42` `folder_breaches`: three paths, `held()` beside `located()`, the checkout's own script only under `same_repo`, and a blob hashed by hand.
- Locality, Self-sufficiency: `bin/larv_review/plan.py:66` `plan_reasons` and `bin/larv_review/owed.py:46` `shortfall`: the rule for when a scoring run is due is split between the 155-line help and the code, and the number owed hangs on the wording `EMPTIED_WORDS`.
- Self-sufficiency: `bin/loop_claude/cycle/judge.py:21` `INTERRUPT_RULES`, with `:196` `quotes_checked`: behaviour held in prompt text and schema descriptions the code can't explain, and prompts that quote `CLAUDE.md` wording.
- Self-sufficiency: `skills/larv-review/SKILL.md:82` "Round cap": dense rules that point at `CLAUDE.md`, `review-findings.md` and the rules files, read three times before the code matched.
- Self-sufficiency: `bin/rule_hook/round_cap.py:26` `earlier_titles`: fails open on the unattended path where git cannot list the branch's titles, and the cap's wording is spelt in SKILL.md, `rules.py:61` and `round_titles.py`.
- Locality, Shape: `bin/chunk_preflight/choices.py:80` `forge_of`: forge identity and forge calls live in two tool packages, so a second forge lands in three.
- Navigation, Shape: `README.md:131` `bin/`: about 63 entries at one level, the shared libraries told apart only by README prose.
- Navigation: `bin/records/__init__.py:1` `records`: utility code named like the judge's records, and quote semantics in `claude_code/transcript.py:85` `verbatim`.
- Shape: `bin/rule_hook/committed_files.py:14` `committed_file_breaches`: a double-reverse dedupe whose choice of copy no comment explains.
- Shape: `bin/model_routing/route.py:28` `levels_fit`, with `route_of`: two rules in one boolean, and type checks packed into one long line.
- Shape: `bin/loop_claude/loop/reload.py:220` `decoded_route`: one dense guard, and renaming a `LoopRun` field stops a running loop.
- Shape: `bin/rule_hook/main.py:116` `bash_breaches`: a 200-character return line joining four kinds of breach.
- Shape: `README.md:3` trial folders: four finished experiments at the root beside the working areas.
- Self-sufficiency: the seats on this run name the decision-log titles that code cites for a constant's reason (`routing: …` in `routing.py`, `ladder.py`, `judge.py`) as the largest cost; state each reason beside its constant.
Scored L 4 because the refactor touches several components that hold known defects, E 2 because review and a manual check see a regression, I 3 because every session runs these tools. The chunk runs a scoring run, since it is a floor item.