Files
claude-config/todo.md
T
lilleman fa34d832aa
CI / tests (pull_request) Successful in 2m54s
Finish item 400 in the plan
2026-11-18 17:23:05 +01:00

85 KiB

todo

Scoring

Score = 100 * V / (2**S + 32)

Bar = 0.1

Next ID = 429

Items

ID Release Blocked by Exempt Item L E I R S Moves Score
428 0.12.0 principle Make each test that skips at run time create the state it needs or fail naming what is missing, so no test in the repo skips at run time. 3 2 1 4 4 code health by 3, 1.1 by 3 4.8
426 0.12.0 Read each epic issue's scores, its blockers and whether the loop filed it through EpicStore, from both backends. 2 2 2 4 3 1 by 5 1.9
427 0.12.0 426 Re-rank Jira's backlog to match an epic's ranking, moving only the issues the loop filed, so people's issues keep their rank. 3 2 3 6 3 1 by 5 1.9
410 0.12.0 427 Run do-epic <EPIC-KEY>: one tmux session per epic, a window per chunk, up to the epic's cap of running chunks and each repo's parallel-chunks, and never in a repo a loop-claude is running chunks in. 3 2 3 6 6 1 by 5 0.8
416 0.12.0 410 Finish an epic's repo item with a bullet under its repo's CHANGELOG.md → ## Unreleased, or with none where the repo's plan is in Jira. 1 1 1 1 2 1 by 5 2.1
413 0.12.0 410 File a finding about an epic goal in the epic and one about a repo goal by that repo's rules, ask the human where unsure which, and give the judge the epic's records beside the repo's. 3 2 3 6 4 1 by 5 1.6
415 0.12.0 410, 413 Finish an epic: run release-review passes on each repo it changed, work what they file, have the human confirm each goal, post a summary comment, and close the epic in Jira last. 3 2 3 6 5 1 by 5 1.2
418 0.12.0 415 Describe do-epic in README.md's tools and setup, with its machine settings and repo-rule.md. 1 2 1 2 2 1 by 5 2.1
420 0.12.0 End a worker that printed its final result and lingers, by killing the background shells it left, after a few minutes, and print each kill with its evidence on loop-claude's terminal. 2 2 2 4 3 1.5 by 6 2.0
409 0.12.0 Split an epic item that changes several repos into one Jira sub-task per repo, each labelled for its repo, with the parent tracking the whole change, when do-epic plans the item. 2 2 3 5 3 1 by 5 1.9
411 0.12.0 410 Park a chunk whose PR waits on a long gate, keeping its worktree, PR and session in a state file that survives a reboot, and resume its session when the gate passes or a reviewer comments. 3 2 3 6 5 1 by 5 1.2
412 0.12.0 410, 411 Print each chunk of an epic, running, parked with its gate and since when, or stopped, each human task and each open interrupt with do-epic <key> --status, and name a parked chunk's gate in its tmux window's name. 2 1 1 2 3 1 by 5 1.9
414 0.12.0 410 DM the human through the notifier when a human-task issue blocks the epic's next item, and poll until it closes. 2 2 2 4 3 1 by 5 1.9
417 0.12.0 410 Run do-epic end to end against a sandbox epic on a real Jira site with a command of its own, outside run-tests. 2 1 1 2 3 1 by 5 1.9
421 0.12.0 Refuse in rule-hook a Bash or Monitor wait that has no end: a loop without a time or count cap, and a pgrep -f or pkill -f whose pattern the same command line contains. 2 2 3 5 3 1.5 by 4 1.3
316 0.12.0 Launch comprehension seats as an agent type of their own under agents/, without CLAUDE.md and the skill and agent listings a general-purpose subagent carries. 2 2 2 4 3 1.4 by 3 1.0
310 0.12.0 Compact a waiting worker's session shortly before its prompt cache expires, only when the human has not answered an interrupt by then. 3 2 3 6 4 1.4 by 3 0.9
422 0.12.0 420 Check every 5 minutes whether a tmux window of the loop looks stuck by other signs than a lingering worker, investigate it with a headless claude, and kill only the process the investigation names. 3 2 3 6 5 1.5 by 4 0.8
365 0.13.0 Have score-outcomes read the finished and dropped items of a plan outside the checkout, from the project repo's history or from Jira's issue changelogs. 3 2 2 5 4 1.6 by 2, 1.1 by 2 1.2
368 0.13.0 Call a release review due, call a release ready to cut, and tell whether a chunk owes a scoring run, for a plan outside the checkout, from the project repo's history or from Jira. 3 2 2 5 4 1.6 by 2, 1.1 by 2 1.2
309 0.13.0 Run a scoring run's second pass on one head with two seats in place of four, once larv-review agreement shows the seats agree with each other more than a run agrees with its rerun. 2 2 2 4 2 1.4 by 3 1.2
419 0.13.0 Run the external auto-reviewer until its review of the PR's head has no findings, and answer each of its threads on the PR. 3 2 3 6 5 1.1 by 5 1.1
366 0.13.0 Count the chunks of a project whose plan sits outside the checkout in the routing figures, from the project repo's history or from Jira. 3 2 2 5 4 1.6 by 2, 1.4 by 2 1.1
339 0.13.0 Lower the reading cost of the files a chunk reads most before its first commit. 2 2 1 3 3 1.4 by 3 1.0
341 0.13.0 Hand each worker's claude a lower Read limit through CLAUDE_CODE_FILE_READ_MAX_OUTPUT_TOKENS, kept only where replays of the archived review rounds find what two unchanged replays find. 2 2 3 5 3 1.4 by 3 1.0
298 0.13.0 Send the judge only the decision entries and the transcript tail a verdict needs, and keep the cut only where judge-replay changes no verdict kind. 3 3 3 7 4 1.4 by 2, 1.5 by 1 0.9
395 0.13.0 decision Keep each block a save didn't change, with its localId, and the spacing paragraphs beside it as Jira or Confluence holds them. 3 2 2 5 5 1.6 by 4 0.8
425 0.13.0 Show on loop-claude's terminal why it doesn't start a chunk beside the next item, as chunk-preflight prints it. 1 1 2 2 2 1.5 by 2 0.7
367 0.13.0 Check that every write to a Jira plan leaves it ranked, as rule-hook's todo-rank-check checks todo.md today. 2 2 3 5 3 1.6 by 2 0.7
423 0.13.0 Skip the judge's model call on a worker's final message that the offline classifier decides is finished. 3 3 3 7 5 1.4 by 2 0.4
77 0.14.0 Teach bin/metric-bench Go and TypeScript, rerun it once two projects' comprehension histories name places, and bring into review only a metric that tracks the panel. 4 2 1 5 6 1.1 by 7, 1.4 by 7 2.1
196 0.14.0 decision Replay the architecture, principles, overrides and product-owner passes with review-replay on Opus 5.5 projects, record each run in the review-replay routing decision, and, where a run qualifies a rung, add back the code that lowers a pass to it. 2 2 3 5 2 1.4 by 3 1.2
321 0.14.0 196 decision Evaluate the review passes that run on a rung review-replay qualified, against the chunks before them, and revert any pass whose findings or later fixes worsened. 1 2 1 2 2 1.1 by 3, 1.4 by 2 2.0
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, Locality first; today the scoring run at 55235a0 scored Navigation 7.38, Locality 6.13, Shape 6.50, Self-sufficiency 6.50 and Overall 6.50. 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
149 0.14.0 decision Fill each release by priority up to a capacity it states, in place of the bar, and move what does not fit to a later release. 4 2 3 7 5 1.1 by 7 1.6
165 0.14.0 Make this repo's tests assert the facts a prompt or help text carries in place of its sentences, and drop the tests that scan source or assert a constant. 3 2 1 4 5 1.1 by 4, 1.4 by 3 1.6
392 0.14.0 owner Group bin/'s packages by concern into seven folders (plan, records, integrations, loop, review, forge and host), so a newcomer finds a tool's neighbours in one place. 3 1 3 5 7 code health by 4 1.6
151 0.14.0 decision Keep 10 to 20 settled decisions per project as its goals' test set, and rerun them after a goal changes. 3 2 2 5 5 1.2 by 7 1.6
68 0.14.0 principle Test workflow-bench/compare.py's blinding and bench.py's run plan. 2 1 1 2 4 1.1 by 5 1.5
102 0.14.0 Give loop-claude's retry delay, and the file it reads the machine's memory from, environment overrides, so the black-box test stops replacing time.sleep and machine_load.memory.MEMINFO through sitecustomize. 1 1 2 2 3 1.1 by 4 1.5
135 0.14.0 Fingerprint only the modules loop-claude imports, so editing another tool's code no longer reloads a running loop. 2 2 3 5 3 1.1 by 4 1.5
5 0.14.0 decision Key a project on its checkout's full path, and refuse to start while another session uses that key. 4 2 3 7 5 1.1 by 6 1.4
291 0.14.0 owner Let principles/technical-principles.md → "Sane defaults" allow a library whose README goals ask that it work with no config to default the clock and randomness in its public constructor, where a documented option overrides each. 1 2 3 4 1 1.2 by 3 1.3
156 0.14.0 decision Drop rank_todo.py's ranking of an old-shape table once no project under ~/code holds one. 2 1 3 4 4 1.1 by 4 1.2
388 0.14.0 Refuse an agent's Write or Edit of a routing ladder file in rule-hook, since a change to the ladder is the human's. 1 1 3 3 2 1.2 by 3 1.2
260 0.14.0 decision Compare the first 20 chunks whose worker got the reading rule with the 20 before them, on review rounds past each pass's first, unfinished verdicts, 🔴, 🟡 and misread findings, dollars and hours per chunk, and drop the rule where any rose. 1 2 1 2 2 1.4 by 3 1.2
51 0.14.0 principle Hand loop-claude's terminal output to an injected effect, so no storage or retry function prints. 2 1 3 4 5 1.1 by 5 1.1
201 0.14.0 owner Save the human's tmux setups and restore them after a reboot, so work is up and running again quickly. 3 2 1 4 4 1.5 by 4 1.1
267 0.14.0 Keep waiting when an interrupt gets an empty answer, at the terminal or in the claude opened for it, in place of stopping the loop. 2 1 3 4 2 1.5 by 3 1.1
4 0.14.0 Name each chunk's archive folder by the chunk's start time, so <key>/ lists chunks in the order they ran. 3 2 3 6 3 1.1 by 3 1.1
324 0.14.0 Count in the routing figures the finished chunks whose worker wrote no Chunk items: line. 2 2 1 3 3 1.1 by 2, 1.4 by 1 1.1
187 0.14.0 Count the tokens a worker spends reading reviewers' reports, so goal 1.4's signal shows whether report files lowered what a review costs the worker's context. 2 2 1 3 3 1.4 by 3 1.0
266 0.14.0 Move the chunks archived under the key home/lilleman/.claude to the key claude-config, so the routing figures and score-outcomes count this repo's chunks as one project. 3 2 2 5 3 1.4 by 3 1.0
181 0.14.0 principle Have the test of a quoted answer's step-up assert the stepped rung's facts as literals, not through prompts.rungs_held. 1 1 1 1 1 1.1 by 2 0.9
193 0.14.0 Narrow numbered's mark to Mark by a typed lookup in place of typing.cast, so a mark added to the regex but not the lookup fails at run time, and one added to the lookup but not Mark fails mypy. 1 1 3 3 1 1.1 by 2 0.9
194 0.14.0 Make test_waits_out_a_usage_limit_which_ends_a_run_of_failures expect a usage-limit wait independent of the wall-clock time that passes during the run. 1 1 1 1 1 1.1 by 2 0.9
265 0.14.0 Archive each wait that costs a chunk wall time, for a usage limit, low memory, a failed worker start or a heavy slot, with its start and end stamps, so goal 1.5's hours split waiting from work. 3 2 2 5 4 1.5 by 3 0.8
169 0.14.0 Have rule-hook's todo-rank-check catch a todo.md that the same Bash call edits and then commits. 3 1 3 5 3 1.1 by 2 0.7
197 0.14.0 Let the human edit what they type at loop-claude's terminal questions with the arrow keys, so Left moves the cursor in place of printing ^[[D. 2 2 3 5 2 1.5 by 1 0.4
1 0.15.0 question Say in README → Install how the transcript archive is backed up, once bender-infra backs up ai-sandbox's data volume. 1 2 1 2 1 0.0
397 0.15.0 owner Script each step that the chunks of releases 0.10.0 to 0.14.0 repeated by hand and a program can compute, filing one item per step, and file this audit again for the next release. 1 2 1 2 2 1.4 by 3, 1.5 by 3 2.3

Details

428. Make each test that skips at run time create the state it needs or fail naming what is missing, so no test in the repo skips at run time.

Follows item 400's bullet in principles/technical-principles.md → the tests principle: "A test doesn't decide at run time whether to check anything: no conditional skip, early return, or catch around its assertions. A test that needs a state creates it, or fails and names what is missing." Stability review found the repo's own tests break it in 13 places, each a skip decided at run time:

  • root-less permission checks, skipIf(os.geteuid() == 0, ...): 3 in bin/loop_claude/test_black_box.py, 3 in bin/loop_claude/test_loop_claude.py, 1 in bin/routing_figures/test_routing_figures.py, 1 in bin/plan_store/test_stores.py;
  • pytest: skipUnless(HAS_PYTEST, ...) on the class Run in bin/run_tests/test_run_tests.py;
  • docker: the docker info skipif on 2 tests in bin/adf_codec/test_adf_codec.py, and needs_docker in bin/plan/test_touches.py;
  • /proc: self.skipTest("no /proc/meminfo ...") in bin/run_heavy/test_run_heavy.py.

The choice per case is the implementer's: for example, run the permission tests as a non-root user or drop the permission check in favour of an injected failure, take the docker or /proc state from an injected fake, or let the test fail naming what is missing. Done when a grep for skipIf, skipUnless, skipTest and mark.skip in the repo's tests finds none.

Touches:

  • bin/loop_claude/test_black_box.py: Cycles.test_exits_1_when_the_transcript_cannot_be_read, Interrupt.test_exits_1_when_the_transcript_cannot_be_read_after_the_interrupt, Models.test_exits_1_when_the_models_file_cannot_be_written
  • bin/loop_claude/test_loop_claude.py: Cycles.test_stops_on_a_session_folder_it_cannot_read_whole, Cycles.test_stops_keeping_the_transcript_when_the_finished_chunks_session_cannot_be_deleted, RoutingCheck.test_says_a_figure_it_cannot_write_and_goes_on
  • bin/routing_figures/test_routing_figures.py: Archive.test_fails_when_it_cannot_keep_the_state (small)
  • bin/plan_store/test_stores.py: FilePlan.test_a_file_it_cannot_write_fails_naming_it (small)
  • bin/run_tests/test_run_tests.py: HAS_PYTEST, Run (small)
  • bin/adf_codec/test_adf_codec.py: test_round_trip_through_the_real_container, test_the_lossless_pair_gives_back_a_jira_description_identical_and_the_portable_pair_does_not
  • bin/plan/test_touches.py: needs_docker, PredictedTest.test_a_list_written_in_markdown_and_read_back_from_a_jira_description_reads_as_written, PredictedTest.test_a_list_typed_in_jiras_editor_reads_as_its_files_and_parts
  • bin/run_heavy/test_run_heavy.py: RunHeavy.test_a_docker_run_without_limits_gets_three_quarters_of_the_memory_split_across_the_slots (small)
  • bin/machine_load/memory.py: read_memory (small)

426. Read each epic issue's scores, its blockers and whether the loop filed it through EpicStore, from both backends.

Today EpicIssue holds none of the three, so do-epic cannot score or order an epic's items (item 408) or tell a person's issue from its own (item 427). The Jira backend reads them as plan_store/jira_epic.py does: the plan-item issue property, its Filed by key, and the "is blocked by" links. The contract tests in contract.py cover both backends.

Touches:

  • bin/do_epic/backend/protocol.py: EpicIssue (small)
  • bin/do_epic/backend/fake.py
  • bin/do_epic/backend/jira.py: FIELDS, JiraEpic.found
  • bin/do_epic/backend/contract.py
  • bin/do_epic/backend/test_jira.py

427. Re-rank Jira's backlog to match an epic's ranking, moving only the issues the loop filed, so people's issues keep their rank.

Split from item 408 (docs/decisions.md → do-epic: drives one Jira epic …). EpicStore gains a rank call, in both backends and the contract tests. Lift JiraPlan.rank_refusal to a module function both stores call. Walk the target order: where the issue at a position isn't the target's, rank the target's just before it; the slots people's issues hold stay theirs. A rank Jira refuses stops the walk and is reported, since a wrongly ranked team backlog takes hours to restore.

Touches:

  • bin/plan_store/jira_epic.py: JiraPlan.rank_refusal (small)
  • bin/do_epic/backend/protocol.py: EpicStore (small)
  • bin/do_epic/backend/fake.py
  • bin/do_epic/backend/jira.py
  • bin/do_epic/backend/contract.py
  • bin/do_epic/ranking.py

410. Run do-epic <EPIC-KEY>: one tmux session per epic, a window per chunk, up to the epic's cap of running chunks and each repo's parallel-chunks, and never in a repo a loop-claude is running chunks in.

The lock that keeps do-epic and loop-claude apart is the per-project starting flock, shared by both (docs/decisions.md → do-epic: drives one Jira epic …). Each chunk routes on its repo's model and ladder, a no-repo chunk on the machine setting's model. loop/chunk_windows.py start_chunks is a place the seats found hard: make it easier to read where it changes. Where a do-epic cycle's worker, judge or human reads the driver's name, they read do-epic, never loop-claude: item 401 left agent_cycle's own text naming loop-claude, since whether the markers that code matches (TYPED_HERE, TRANSCRIPT_HEADING) vary by driver depends on what do-epic calls itself and where it archives its cycles. The worker's prompt and the judge's rules share the phrase a worker asks for a higher rung with, so both take it from one place.

Touches:

  • bin/loop_claude/loop/chunk_windows.py: starting, start_chunks (small)
  • bin/agent_cycle/tmux.py: session_name (small)
  • bin/do-epic
  • bin/do_epic/main.py
  • bin/do_epic/scheduler.py
  • bin/agent_cycle/worker_prompts.py: TYPED_HERE, stepped_to (small)
  • bin/agent_cycle/judge.py: TRANSCRIPT_HEADING, Reply.escalation (small)
  • bin/agent_cycle/escalation_mark.py: STEP_LINE (small)
  • bin/agent_cycle/interrupt.py
  • bin/agent_cycle/stop.py

416. Finish an epic's repo item with a bullet under its repo's CHANGELOG.md → ## Unreleased, or with none where the repo's plan is in Jira.

An epic item has no release (docs/decisions.md → do-epic: drives one Jira epic …). do-epic runs chunk-pr merge --finish in the repo's checkout, and picks the flag by the repo's plan type.

Touches:

  • bin/do_epic/main.py

413. File a finding about an epic goal in the epic and one about a repo goal by that repo's rules, ask the human where unsure which, and give the judge the epic's records beside the repo's.

The epic's records are its description's sections (docs/decisions.md → do-epic: drives one Jira epic …). read_records and read_premises take the epic's record as a parameter, so loop-claude's judge is unchanged.

Touches:

  • bin/loop_claude/project/records.py: read_records (small)
  • bin/loop_claude/project/premises.py: read_premises (small)
  • bin/agent_cycle/judge.py: judge_prompt (small)
  • bin/loop_claude/cycle/prompts.py: ANSWER_DESTINATIONS (small)
  • skills/larv-review/SKILL.md: ## 3. Passes (small)
  • bin/do_epic/findings.py

415. Finish an epic: run release-review passes on each repo it changed, work what they file, have the human confirm each goal, post a summary comment, and close the epic in Jira last.

A goal is met only where the human confirms it: each achieve goal achieved, and no defect against a maintain goal. For a goal that isn't met, do-epic proposes a set of items, and the human approves the set before work starts (docs/decisions.md → do-epic: drives one Jira epic …). The summary names what shipped per repo and each goal's evidence, in the epic's language. The epic's transition is configured as todo-item finish's is.

Touches:

  • bin/plan_store/jira_epic.py: JiraPlan.finish, transition_of (small)
  • bin/release_plan/release_standing.py: release_reviewed (small)
  • bin/do_epic/finish.py

418. Describe do-epic in README.md's tools and setup, with its machine settings and repo-rule.md.

Each chunk keeps the README true for what it changes; this item writes the overview a new reader needs (docs/decisions.md → do-epic: drives one Jira epic …).

Touches:

  • README.md: ## Tools, ### Set up a project (small)

420. End a worker that printed its final result and lingers, by killing the background shells it left, after a few minutes, and print each kill with its evidence on loop-claude's terminal.

On 2026-11-18 the worker of the item 402 chunk merged its pull request at about 11:47, printed its summary and then waited 95 minutes: it had started until [ -s /tmp/merge-402.log ] && ! pgrep -f "chunk-pr merg"; do sleep 15; done in the background, and pgrep -f matched the loop's own command line, so the condition never held. Its claude -p stayed alive ("Waiting for background work to finish"), loop-claude waited for it to exit, and the window's slot started no new chunk until the human found it and killed the shell.

Every 5 minutes loop-claude checks each worker it runs: when the worker's last transcript entry is its final result and its claude -p is still alive more than 3 minutes later, it kills the background shells that are children of that process, and the worker then exits. Nothing else is killed: no worker that is still working, no process of another session, and no window with a sign short of this one. A worker whose shell is still writing output or holds a run-heavy slot is left alone. Each kill is printed on loop-claude's terminal with the process, its age and its command, and appended to the judge's records so the pattern can become a rule-hook ban (item 421). Item 422 handles the vaguer signs of a stuck window.

The check lives in the main run beside the other per-cycle work; item 401 moves that code into a package, and a conflict with it is cheaper than waiting for it.

Before building the check, try the platform's own bound: Claude Code documents BASH_MAX_TIMEOUT_MS and BASH_DEFAULT_TIMEOUT_MS as the time limit of a background command, 10 minutes by default in claude -p (docs: code.claude.com/docs/en/headless.md and env-vars). Nothing in this repo sets them, yet the lingering shell ran 95 minutes on Claude Code 2.1.327, so first find out why the limit did not apply (an explicit timeout in the worker's call, or a gap in the docs). Hand the workers a cap through those variables where that bounds the wait, and build the lingering-worker check only for what the cap leaves.

Touches:

  • bin/loop_claude/loop/main.py: run_loop (small)
  • bin/loop_claude/loop/lingering_workers.py
  • README.md: ## Tools (small)

409. Split an epic item that changes several repos into one Jira sub-task per repo, each labelled for its repo, with the parent tracking the whole change, when do-epic plans the item.

do-epic predicts the repos from the code, as Touches: predictions are made. A person's issue is split too, with a comment for the team in the epic's language (docs/decisions.md → do-epic: drives one Jira epic …). The parent is finished once every sub-task is.

Touches:

  • bin/plan_store/jira_epic.py: Filing, filing_of (small)
  • bin/do_epic/split.py

411. Park a chunk whose PR waits on a long gate, keeping its worktree, PR and session in a state file that survives a reboot, and resume its session when the gate passes or a reviewer comments.

A parked chunk takes no running slot, and the items it blocks wait (docs/decisions.md → do-epic: drives one Jira epic …). On reviewer feedback the resumed chunk fixes what is reasonable, replies to the rest, and changes the setup so review would have caught it. The forge checks need forms that return instead of blocking. chunk_pr/main.py merged is a place the seats found hard: make it easier to read where it changes.

Touches:

  • bin/chunk_pr/main.py: merged, merge_turn (small)
  • bin/chunk_pr/approval.py: approval, unanswered (small)
  • bin/forge/pulls.py: checks_passed (small)
  • bin/do_epic/gates.py
  • bin/do_epic/parked.py

412. Print each chunk of an epic, running, parked with its gate and since when, or stopped, each human task and each open interrupt with do-epic <key> --status, and name a parked chunk's gate in its tmux window's name.

A plain do-epic <key> resumes all of them, as after a reboot (docs/decisions.md → do-epic: drives one Jira epic …). The window suffix generalises the interrupt's " ?".

Touches:

  • bin/agent_cycle/tmux.py: WAITS, mark (small)
  • bin/do_epic/status.py

414. DM the human through the notifier when a human-task issue blocks the epic's next item, and poll until it closes.

do-epic leaves the issue as the team filed it (docs/decisions.md → do-epic: drives one Jira epic …).

Touches:

  • bin/agent_cycle/direct_message.py: tell (small)
  • bin/do_epic/human_tasks.py

417. Run do-epic end to end against a sandbox epic on a real Jira site with a command of its own, outside run-tests.

The sandbox is the Jira site larvit.atlassian.net, project SCRUM: a team-managed test-data space with Epic, Task, Story and Subtask issue types. The test creates its own epic there, with children and sub-tasks, and deletes them when it ends, so it leaves the space's other issues alone. Its account may create, edit, transition, comment on and delete issues in SCRUM. The command reads the env-file's path, holding ATLASSIAN_EMAIL and ATLASSIAN_API_TOKEN, from a setting of its own, never from the repo.

The test never decides at run time whether to check anything: the command fails, naming what is missing, where the site or its env-file isn't set (docs/decisions.md → do-epic: drives one Jira epic …).

Touches:

  • README.md: ## Test (small)
  • bin/do-epic-e2e
  • bin/do_epic/e2e/

421. Refuse in rule-hook a Bash or Monitor wait that has no end: a loop without a time or count cap, and a pgrep -f or pkill -f whose pattern the same command line contains.

On 2026-11-18 a worker in the item 402 chunk started until [ -s /tmp/merge-402.log ] && ! pgrep -f "chunk-pr merg"; do sleep 15; done in the background. The pgrep -f pattern is part of the loop's own command line, so it matched itself and the loop never ended; the worker's claude -p waited on it for 95 minutes and its slot started no new chunk (item 420 ends such a worker; item 422 handles other stuck windows). A wait that cannot end is cheaper to refuse before it starts.

rule-hook refuses a Bash command, or a Monitor call, that starts a wait of either kind. A loop (until, while, for) that sleeps and has neither a time limit (timeout, a deadline compared with date) nor a bounded count (for i in $(seq …)); ~/.claude/bin/wait-for always passes, since it takes --timeout. A pgrep -f or pkill -f whose pattern text also appears in the command line that holds it, quoted or not, so it can match its own shell. The refusal names which kind it found and says to use wait-for <file> [<pattern>] --timeout <seconds>, or a pgrep -x on the process name.

settings.json hooks only Agent|Bash|Edit|Grep|MultiEdit|Read|Skill|Write, so rule-hook never sees a Monitor call: add Monitor to that matcher, and have rule-hook read a Monitor call's command as it reads a Bash one. Claude Code 2.1.314 and later send Monitor-started commands to a PreToolUse hook; a hook's allow and updatedInput are ignored for them, so rule-hook can refuse but not rewrite (docs: code.claude.com/docs/en/hooks.md).

The test covers each kind refused, a bounded loop that passes, wait-for passing, and a pgrep -f on a pattern absent from the command passing. The rule's reason goes in rule-hook's rule table beside the other rules.

Touches:

  • bin/rule_hook/rules.py: RULES (small)
  • bin/rule_hook/endless_wait.py
  • bin/rule_hook/test_rule_hook.py: CommandBreaches (small)
  • settings.json

316. Launch comprehension seats as an agent type of their own under agents/, without CLAUDE.md and the skill and agent listings a general-purpose subagent carries.

Seats run as general-purpose subagents, so each starts with the skill and agent listings, about 7,500 tokens, at $9.00 per 1,000 prefix tokens over the archive measured from 2026-09-30 to 2026-11-11: about 1.3% of spend. An agent under agents/ with Read, Grep, Glob and Bash, omitClaudeMd: true, and the seat's persona in its prompt drops both; a seat reads the project as a newcomer does, which is what it simulates. Keep each seat's prompt from skills/comprehension-panel/seat-prompts.md. Check that a scoring run's scores on one head stay within the spread two runs of today's seats show. lilleman approved on 2026-11-11.

Touches:

  • agents/comprehension-seat.md
  • bin/rule_hook/reviewer_rung.py: SEAT, review_kind
  • bin/larv_review/scoring_setup.py: SEATS, prepare_scoring
  • bin/loop_claude/cycle/prompts.py: review_launches (small)
  • skills/comprehension-panel/seat-prompts.md: ## Opening, (top) (small)
  • skills/comprehension-panel/SKILL.md: ## The scoring run (small)
  • skills/larv-review/SKILL.md: ### Comprehension check (small)
  • bin/rule_hook/test_rule_hook.py: ReviewerRung
  • bin/larv_review/test_scoring_setup.py

310. Compact a waiting worker's session shortly before its prompt cache expires, only when the human has not answered an interrupt by then.

281 worker resumes after the 1-hour cache expired rewrote a whole context, $379 of the $9,154 measured from 2026-09-30 to 2026-11-11 (4.1%). A compaction while the cache is warm reads the context at a twentieth of the input price and leaves about 20k tokens to rewrite later.

lilleman, 2026-11-11: do it on a timer, so an answer that comes in time doesn't cost anything. When loop-claude interrupts, start a timer a margin short of the cache lifetime (about 50 minutes of a 1-hour cache; read the lifetime from the session's last cache_creation usage). If no answer has come when it fires, compact the session, then keep waiting. Skip it below about 60k tokens of context, where the rewrite costs less than the compaction. A wait for a usage limit can't be helped this way.

Probe first that claude -p "/compact" --resume <id> compacts a session: the Agent SDK documents /compact as a prompt, the CLI pages don't show it. Claude Code compacts idle interactive sessions itself ("Compacted while idle"), but not -p runs. Check: the cost entries of 20 chunks with an interrupt, before and after.

Touches:

  • bin/agent_cycle/interrupt.py: interrupt, answered, typed_answer, WorkerSession
  • bin/loop_effects/effects.py: Effects, system_effects
  • bin/claude_code/tokens.py: tokens_of (small)
  • bin/claude_code/reply.py: claude_reply (small)
  • bin/loop_claude/loop/usage.py: HELP (small)
  • bin/loop_claude/test_loop_claude.py: FakeEffects, Interrupts
  • bin/loop_effects/test_effects.py

422. Check every 5 minutes whether a tmux window of the loop looks stuck by other signs than a lingering worker, investigate it with a headless claude, and kill only the process the investigation names.

Item 420 ends a worker that printed its final result and lingers. A window can also stall earlier: a test run that hangs on a shell ignoring SIGINT (item 402's first full run), a pane with no output while its process tree does no work, or a background shell older than the chunk's usual length.

Every 5 minutes loop-claude lists its own session's windows and, for one with such a sign, launches a headless claude on the project's rung that reads the pane, the process tree and the chunk's log files and returns working, waiting on a person or stuck, with the process that holds the window and why. Only stuck kills, and only the named process, never the worker or the loop. A window waiting on a gate (run-heavy, a test run that still writes output) or on the human's answer is not stuck. Each check and kill is printed on loop-claude's terminal with the evidence.

Run it first as a report that kills nothing, and turn the kill on once its verdicts over a few days match what the human would have done; record that in the item's decision. Blocked by 401, which moves loop-claude's cycle into a package; this item lands in the package.

Touches:

  • bin/loop_claude/loop/stuck_windows.py
  • bin/agent_cycle/tmux.py: Listed, listed
  • README.md: ## Tools (small)

365. Have score-outcomes read the finished and dropped items of a plan outside the checkout, from the project repo's history or from Jira's issue changelogs.

lilleman's call, 2026-11-12: 0.10.0 and 0.11.0 switch the history tools off for a plan outside the checkout, and 0.13.0 holds only their ports. In Jira a dropped item, which Drop item commit titles mark today, is an issue resolved by a resolution the config names.

368. Call a release review due, call a release ready to cut, and tell whether a chunk owes a scoring run, for a plan outside the checkout, from the project repo's history or from Jira.

One of 0.13.0's history ports, per lilleman's call of 2026-11-12. Filed from item 344's stability review, round 1.

Today a plan outside the checkout is read as it stands, since no commit holds it. So:

  • bin/release_plan/release_standing.py's todos_since holds one text. Once a release's last item ships, the plan no longer names the release, and next_release gives the release after it as next. chunk-preflight and loop-claude say UNSEEN_RELEASES on every run: the release reviews and the cuts are the human's to call.
  • bin/larv_review/owed.py's owed_scoring_run says it cannot tell whether the chunk owes a scoring run, and the merge goes on. needs-scoring-run and record-scoring refuse the ID of an item the plan no longer holds.

Read the plan's earlier texts from the project repo's commits, which item 345 writes, or the releases Jira's issues name. Then drop UNSEEN_RELEASES and both refusals, and the bullets that record them in docs/decisions.md → "project config: a tool that reads a file's commits …".

The comprehension seats found next_release hard to read: the tag cases, dated headings, renames and named_releases meet in one function. Make it easier to read while there.

309. Run a scoring run's second pass on one head with two seats in place of four, once larv-review agreement shows the seats agree with each other more than a run agrees with its rerun.

Seats cost $1,143 of the $9,154 measured from 2026-09-30 to 2026-11-11 (12.5%), scoring runs $470 of that. A scoring run today is four seats, run twice on one head: eight seat runs. Two seats on the second pass make it six, and two on both passes four.

Why it should hold quality: the seats are one model with different personas, so their errors are correlated, and a correlated extra seat adds little. larv-review agreement showed on 2026-11-11 that Mid A and the inherited architect voted alike in 58 of 75 paired rulings, while their scores moved alike between two runs on one commit in only 1 of 4 pairs: seats agree with each other more than a run agrees with its rerun, so the rerun carries more new information than the third and fourth seat. Nine judges from seven model families carried about two independent votes (arXiv 2605.29800), and ten with error correlation 0.21 carried 3.5 (arXiv 2609.22512).

Decide once larv-review agreement prints Assumption holds: or Assumption fails:; it needs 10 commits each scored twice before it can. Keep four seats on any pass that files a floor item, since its places come from seat reports. Revise the decision review: the comprehension check records how hard a project is to read…, which names the four-seat run.

lilleman moved it from 0.12.0 to 0.13.0 on 2026-11-17, when 6 commits had been scored twice.

Touches:

  • bin/larv_review/seats.py: SCORING_SEATS, scores
  • bin/larv_review/history_commands.py: record_scoring
  • bin/larv_review/scoring_setup.py: SEATS, prepare_scoring
  • bin/larv_review/main.py: (module) (small)
  • bin/larv_review/test_larv_review.py
  • bin/larv_review/test_scoring_setup.py
  • skills/larv-review/SKILL.md: ### Comprehension check
  • skills/comprehension-panel/SKILL.md: ## The scoring run (small)

419. Run the external auto-reviewer until its review of the PR's head has no findings, and answer each of its threads on the PR.

Today larv-review §3 → "Round cap" stops the auto-reviewer after two reviews. On a client project's pull request, CodeRabbit spent both reviews on the draft, before the internal passes were done, so its last finding got no answer on the PR. Change:

  • §3 → "Round cap": drop the auto-reviewer's cap. Every finding it makes, whatever its severity, is fixed or shown wrong, and each push waits for its review of the new head. The pass is done when that review adds no finding.
  • §4 step 6: detect a clean review. CodeRabbit posts no review when it has no findings. Instead it updates its summary comment with "No actionable comments were generated in the recent review" and a coveredCommitId that matches the head. Wait for that, not for a new review.
  • Answer every thread in place, with the fix's commit or the reason for not making it.
  • chunk-pr ready and chunk-pr merge refuse while an auto-reviewer thread has neither our reply nor resolved status, read from the thread state GitHub's API reports.
  • round-count, rule-hook's round-cap and the "Answer … round" titles need no change, since the auto-reviewer's commits already take ordinary titles.
  • Once this ships, delete the first two bullets of the machine-local CodeRabbit rule file, which say the same thing for one machine.

The change touches skills/larv-review/SKILL.md "Re-entry", which the comprehension seats found hard to follow (Locality): make it easier to read while editing it.

366. Count the chunks of a project whose plan sits outside the checkout in the routing figures, from the project repo's history or from Jira.

One of 0.13.0's history ports, per lilleman's call of 2026-11-12.

339. Lower the reading cost of the files a chunk reads most before its first commit.

reading-cost printed Rising: on 2026-11-12: the last 5 chunks' median is 128,199 new input tokens before their first commit, the 5 before them 101,795. Read most over the last 5, as opens and searches: todo.md 34, docs/decisions.md 26, README.md 19, skills/larv-review/SKILL.md 13, bin/agent_cycle/routing.py 8, CHANGELOG.md 6. The chunk that ships this runs reading-cost --reset.

Touches:

  • README.md
  • skills/larv-review/SKILL.md
  • bin/agent_cycle/routing.py

341. Hand each worker's claude a lower Read limit through CLAUDE_CODE_FILE_READ_MAX_OUTPUT_TOKENS, kept only where replays of the archived review rounds find what two unchanged replays find.

Raised by item 312's research round, 2026-11-12. Over the 53 chunks archived since 2026-09-30, Read results held about 3.17M tokens (characters / 4) in 1,617 reads. Results over 20,000 characters, about 5,000 tokens, held 479k tokens beyond that size in 115 reads: prose reviewers 144k, stability reviewers 119k, product-owner reviewers 102k, general-purpose subagents 65k, principles auditors 28k, architects 21k. Workers had none, under the reading rule in their prompt. Measured with a script over loop-claude's archive that sums each Read tool_result's characters by the subagent's agentType. The decision of 2026-10-27 found Read results out of a PostToolUse hook's reach; this variable reaches them.

CLAUDE_CODE_FILE_READ_MAX_OUTPUT_TOKENS overrides the Read tool's default limit of 25,000 tokens (https://code.claude.com/docs/en/env-vars, read 2026-11-12: "Override the default token limit for file reads, which is 25,000 tokens."). Claude Code 2.1.322's binary holds the name. A read over the limit fails and asks for an offset and a limit, so an agent may page through the same file and save nothing; the replay measures that.

loop-claude can hand it as it hands CLAUDE_CODE_AUTO_COMPACT_WINDOW, and subagents inherit it, the principles auditors included. The choice of 2026-10-27 that the principles audit reads the project whole holds as long as it pages through what it reads.

The replay: review-replay <pass> claude-opus-5-5 --read-limit already replays a pass's archived rounds three times as archived and once at 5,000 tokens (bin/review_replay/changes.py, bin/claude_code/output_limits.py). Run it for the prose, stability, product-owner and principles rounds. Keeps quality where each pass finds every finding the two unchanged replays both find, or misses no more than the most an unchanged replay misses, as --omit-claude-md counts it. Keep it only where the replays' Read characters also fall; the archived runs hold each result. Then hand the variable over in worker_environment for what passes.

Touches:

  • bin/agent_cycle/run.py: worker_environment (small)
  • bin/loop_claude/loop/usage.py: HELP (small)
  • bin/loop_claude/test_loop_claude.py: Cycles (small), HelpLimits (small)

298. Send the judge only the decision entries and the transcript tail a verdict needs, and keep the cut only where judge-replay changes no verdict kind.

153 Judge cost: entries total $109.76, a median of about $0.76 a verdict at every rung, and the cost follows the size of the records, not the transcript: about 76k characters in log and 188k in claude-config. In 224 of 251 recent verdicts (89%) the evidence sits in the worker's final message. A missed interrupt is the costly error, so the bar is zero changed interrupts and zero new rejections across the replay. The judge is about 4.5% of dollars, so the saving is small; each cycle also gets faster.

Touches:

  • bin/agent_cycle/judge.py: Brief, judge_prompt, JUDGE_PROMPT_CHARS
  • bin/loop_claude/project/records.py: read_records
  • bin/judge_replay/main.py: rule (small)
  • bin/judge_replay/replay.py: case_from (small)
  • bin/loop_claude/test_loop_claude.py: JudgePrompt, Records
  • bin/judge_replay/test_judge_replay.py

395. Keep each block a save didn't change, with its localId, and the spacing paragraphs beside it as Jira or Confluence holds them.

Today a Jira detail edited in one line, or a Confluence record with one entry appended, is written back whole by adf-codec's portable pair, which drops each block's localId and the empty paragraphs Jira uses for spacing. A save whose text is the text a read showed already sends nothing (docs/decisions.md → project config: the loop shows a Jira description and a Confluence page as adf-codec's portable markdown, …). Match the saved markdown's top-level blocks against the held document's, each as the portable pair reads it, and send the held block where they match. The decision's limit on partial saves ends with this item.

425. Show on loop-claude's terminal why it doesn't start a chunk beside the next item, as chunk-preflight prints it.

chunks_beside names each item it refuses and why, such as an unreadable Touches: list, and chunk-preflight prints those lines, but loop-claude's chunk_windows.may_start keeps only picked and drops them, so idle slots show no reason. Found by item 424's product-owner review; a list a person edits in Jira's UI, or one filed before item 424, still reaches loop-claude unchecked.

367. Check that every write to a Jira plan leaves it ranked, as rule-hook's todo-rank-check checks todo.md today.

One of 0.13.0's history ports, per lilleman's call of 2026-11-12.

423. Skip the judge's model call on a worker's final message that the offline classifier decides is finished.

On 2026-11-18 lilleman ruled that the classifier's miss rate of about 1 in 100 decided is close enough to no missed interrupt (docs/decisions.md → routing: the offline classifier may decide finished, since lilleman ruled a miss rate of about 1 in 100 decided close enough to none; the judge's model keeps every other verdict kind).

Taking the item builds this: before the judge call, loop-claude runs the classifier trained on its archive over the worker's final message. Where the classifier decides finished, loop-claude skips the call. The classifier is re-trained as the archive grows. The judge call sits in bin/agent_cycle/run.py → rule, a place the comprehension seats found hard, since the order of stamping, archiving and stepping matters across three functions: make it easier to read.

The judge's model still decides every verdict the classifier leaves, the escalation a worker quotes, and whether a release is ready to cut, since the replay didn't test the classifier on those.

Evidence, from judge-replay claude-opus-5-5 --offline: the classifier reads only the final message, averages five seeded logistic regressions, and decides finished only where its chance of an interrupt is below that of every interrupt among its held-out training folds.

  • On the 1,116 archived verdicts it decided finished with no missed interrupt on both tests: held-out folds, 95 decided and none missed; latest fifth, 20 decided and none missed.
  • On the first 1,104 it missed one interrupt on the latest fifth, while the held-out folds decided 103 and missed none.
  • It decides about 1 in 12 verdicts. About 1 in 100 of those it decided were an interrupt, on the 1,104.

Savings: one judge call per about 12 verdicts. That is small, and nobody has measured it.

77. Teach bin/metric-bench Go and TypeScript, rerun it once two projects' comprehension histories name places, and bring into review only a metric that tracks the panel.

2026-10-08, lilleman: no metric enters review until a rerun on more data shows it tracking the panel.

Before the rerun, set the AUC and top-k lift over chance that count as tracking, from a published threshold, so the rerun ends in a decision. The rerun needs places from a project in Go or TypeScript, the languages most other projects use, so the bench first measures those languages, run through each project's own tooling. Function and file length stay warnings to look into, never limits: a plain long function can be simple, and a hard one may need five-line pieces.

On 2026-10-08 only this repo's history named places: five scoring runs, all that day. The medians:

Metric AUC Top k
Cognitive complexity 0.66 0.26
Length 0.74 0.18
Shared state constant constant
Change spread 0.67 0.18

AUC is 0.5 for a coin toss, and top k scores 0.10 by chance. Shared state cannot be judged here: every unit in bin/ scores 0 on it. Every run's first place was a section of skills/larv-review/SKILL.md, which no code metric sees.

On 2026-10-09 a question bench asked a headless claude, allowed only Read, Grep and Glob, three questions a script answers from the syntax tree: which units call a unit, which units it calls, and which test files refer to it. It asked them about every unit the history names whose name is unique, plus as many units the history does not name. Five of the history's six scoring runs named such a unit. The tool is in commit 083cc78; it was deleted after the run.

Claude's misses ranked the named units at a coin toss. Top k scores 0.50 by chance on these units:

Metric AUC Top k
Question misses 0.49 0.50
Cognitive complexity 0.66 0.59
Length 0.75 0.69
Change spread 0.57 0.54

Claude's answers overlapped the truth by a median Jaccard index of 0.76. A run took in 62,643 new input tokens and cost $0.70; it did not record which model answered. Grep answers these questions whatever the code's shape, so no question enters review; a new question bench would need questions that take understanding the code, not finding it.

196. Replay the architecture, principles, overrides and product-owner passes with review-replay on Opus 5.5 projects, record each run in the review-replay routing decision, and, where a run qualifies a rung, add back the code that lowers a pass to it.

Item 159's chunk, PR #322, replayed only the prose and stability passes. Sonnet 5.5 at high missed 44 to 55% of what the project's rung found again in each, so the agent judged a replay of the other four unlikely to qualify a rung. The product-owner review of PR #322 asked that the plan keep them, so the ranking decides when they run. Each run costs hours of reviewer replays (goal 1.4), so the score keeps it low until a cheaper model joins the ladder. docs/decisions.md → routing: a review pass runs below the project's rung only on a rung whose replay of its rounds finds a share of their 🔴 and 🟡 findings no more than a margin below the project's rung's own replay… records the runs so far. The release review of 0.6.0 deleted the code that lowered a pass, since no rung had qualified; the decision says where it goes back.

321. Evaluate the review passes that run on a rung review-replay qualified, against the chunks before them, and revert any pass whose findings or later fixes worsened.

The prose pass's replay under item 318's new bar, on 2026-11-12, qualified no rung, so this item waits for item 196's replays of the other passes. lilleman, 2026-11-11: that bar is a risk, so check it closely a release later. Compare the first 20 chunks reviewed on each newly qualified rung with the 20 before: review rounds past the first, 🔴 and 🟡 per chunk, misread findings, and later fixes that score-outcomes counts (defects that escaped review). Revert a pass to the project's rung where any of those worsened beyond the spread of the 20 before.

273. 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.

docs/decisions.md → routing: one ladder… records why lilleman kept the steps on 2026-11-08. Of the 36 chunks it measured, 1 scored an E + I of 6 or more and none 7 or more, so the steps for 6 to 8 are untested. Start once score-outcomes shows 10 finished chunks with an E + I of 6 or more. Then read its rank correlation for E + I, and the routed-chunks table of routing-figures.md in loop-claude's archive ($LOOP_CLAUDE_TRANSCRIPTS, default ~/.claude/loop-claude-transcripts). Propose new steps to lilleman, whose call they are.

268. 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.

On 2026-11-08 the human answered notice only: the loop prints that the release is ready to cut and ends; no interrupt, and nothing waits in loop-inbox.

Releases are the human's, and CLAUDE.md → "Working loop" already ends the loop once a release is ready to cut, so the ask settles nothing. 15 interrupts in the archive ask whether the human will cut a release, and about 6 of the 45 interrupts from 2026-11-02 to 2026-11-10 said a release was ready to cut; log's 2.5.0 was asked twice within an hour on 2026-11-10. On 2026-11-08 one of the 4 questions waiting in ~/.local/state/loop-claude/ is such an ask, plainpages-plugin-fastjira's 0.3.0 since 2026-11-03. Today a release ready to cut opens claude's session at once (bin/agent_cycle/run.py ask), and the loop stops if the human leaves it with nothing typed. The change also touches what an interrupt covers in CLAUDE.md → "Vocabulary". Where the change touches ask, make it easier to read, as the seats named it.

171. 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.

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, Locality first; today the scoring run at 55235a0 scored Navigation 7.38, Locality 6.13, Shape 6.50, Self-sufficiency 6.50 and Overall 6.50.

Filed by the comprehension check of PR #473 at 13d1550 (larv-review record-scoring), which printed Floor: and Next:, and refreshed from the two runs of PR #560 at 55235a0. The lowest mean is Locality, 6.13 in both runs: the seats name rules held partly in agent-facing prose and partly in code, and state that one unit can't be read without three others open. Self-sufficiency and Shape follow at 6.50. 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 first:

  • Locality, Self-sufficiency: bin/release_plan/release_standing.py:176 next_release, with :162 carried: the cut from a tag, a dated heading or BEFORE_FIRST_RELEASE, release names replayed from the plan's history, and renames meet in the is_earliest test at :198; carried is reapplied in chunk_item and in chunk_preflight/choices.py planned, one relying on zip keeping order. All eight seats named it; four would least want to change it.
  • Locality: bin/agent_cycle/routing.py:116 unfinished_route, with :131 escalated_route: five rungs, the unfinished count and the escalation mark, which spans escalation_mark.py, judge.py, verdicts.py, route.py and reload.py. Seven seats named it; the module table is what makes it readable.
  • Locality: bin/chunk_pr/main.py:298 merged, with :384 levelled_and_finished and :407 finished: a loop around the queue turn binds done only under some if ready branches, the approval wait is a fall-through, and finished destructures five subprocess results by position.
  • Locality: skills/larv-review/SKILL.md:82 "Round cap" and :90 "Re-entry", with bin/review_rules/round_titles.py:45 and :63 past_cap, bin/rule_hook/round_cap.py:10 and bin/larv_review/reviewer_header.py:73: the round cap lives in five places, and SKILL.md's Done, Cap, Re-entry and Mechanical answers point at each other as prose.
  • Locality: bin/loop_claude/loop/chunk_windows.py:107 start_chunks: under one flock it checks the session's owner, splits running from stopped windows, computes the room, reads the plan with a fallback, gathers held items and filters again, with the limit applied a second time in may_start.
  • Shape, Locality: bin/release_plan/ranking.py:239 score, :144 weigh and :250 weight_of: every function forks on SHAPE versus OLD_SHAPE, and one Score depends on five inputs held in other files.
  • Locality, Shape: bin/plan_store/stores.py:1 and bin/release_plan/todo_table.py:1: release_plan and plan_store import each other, the plan has three names, and the goals file sits under release_plan.
  • Shape: bin/larv_review/plan.py:22 scoring_run_due, with owed.py, unmeasured.py and scoring.py: the scoring-run trigger is spread over four modules, three of whose names don't say it, in a flat 14-module package that mixes rounds, the comprehension check and its triggers.
  • Self-sufficiency: bin/agent_cycle/judge.py:22 INTERRUPT_RULES, with :197 quotes_checked: the judge's behaviour is held partly in prompt text and field metadata and partly in Python checks.
  • Shape: bin/plan_store/jira_epic.py:356 revision_of: indexes the untyped tuple scores at :530 returns by position.
  • Shape: bin/agent_cycle/run.py:55 Cycle.todo, bin/agent_cycle/hand_over.py:23 hand_over and bin/release_plan/named_code.py:5: names that say todo.md where the plan may be a Jira epic, a handover that is not the reload's, and a module with no code in it.
  • Navigation, Shape: README.md:197 the bin/ package list: 74 entries at one level, with the README's list as the only index.
  • Self-sufficiency: every seat names the decision-log titles that code cites for a reason (routing.py:1, ranking.py:20, loop/routing_check.py:1) as a cost; state each reason beside its constant and keep the citation as a footnote.

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.

264. Report the comprehension seats and the escalated subagents as subagent types of their own in routing-figures.md, with their dollars.

Seats launch as general-purpose, so they hide inside that type: $301.24 of the $962.17 the figures of 2026-11-04 count, from 221 subagents. By 2026-11-10 general-purpose was the second-largest spend, $585 of about $2,385, and by transcript size about two thirds of it were seats (291 launches), then 177 security passes. Escalations are counted by trigger only, never in dollars. bin/routing_figures/chunk.py groups subagents by their subagent_type; a seat is told apart by its prompt, and an escalated subagent by the escalation marker its prompt opens with.

323. Count a release review as landed once its pull request has merged, whatever its commits are titled.

release_reviewed in bin/release_plan/release_standing.py counts a release review only by a commit titled Release review of <release> on the default branch. In a replay of archived interrupts (judge-field-trials/), in two projects whose release review had merged, the default branch held no such commit, so chunk-preflight and routing read ReviewDue, and the next loop would run the release review again. skills/larv-review/SKILL.md opens the review's branch with an empty commit of that title, but a worker may skip it, and a rebase or squash merge, which chunk-pr uses where the forge allows no merge commit, may drop it. Detect the review by something every merge method keeps, and check the fix against both projects' histories. Until then the judge keeps release_ready (docs/decisions.md → routing: loop-claude reads the items a chunk takes …).

342. 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.

The next round of item 312's research, filed when 312's round ran on 2026-11-12. lilleman, 2026-11-11: repeat the round once per claude-config release, not per project. When this item's chunk finishes, file the next round into the next claude-config release with items left.

Run it as 312's ran: a lead on the project's rung plans 6 to 10 searches where the archive's spend is, cheaper subagents with web access only return sourced findings, and the lead checks each claim against the archive's measured spend before filing.

Sources that proved useful in 312's round: Claude Code's environment variable reference (https://code.claude.com/docs/en/env-vars), its subagent reference (https://code.claude.com/docs/en/sub-agents), its CHANGELOG (https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md), the API's context editing and tool search pages (https://platform.claude.com/docs/en/build-with-claude/context-editing, https://platform.claude.com/docs/en/agents-and-tools/tool-search-tool), and Anthropic's post on code execution with MCP (https://www.anthropic.com/engineering/code-execution-with-mcp). Check a variable a source names with a grep of the installed Claude Code binary; issue mirrors and blogs gave leads but no figure to rely on. The archive counts that decided: command-patterns <version> --calls, and scripts in /tmp over loop-claude's archive summing tool results by agent type, cache writes by lifetime and request gaps, and each subagent's first request.

Dropped in 312's round, and why; look again only where the reason has changed:

  • Editing through Edit in place of python3 heredocs: Edit already accepts a file read with sed or cat on Claude Code 2.1.319 to 2.1.322, 27 such Edits in 0.7.0 and 0.8.0 and no "not read yet" refusal in the archive, and an Edit would have saved about 2,300 command tokens per chunk.
  • Forcing 5-minute caches on the worker or 1-hour ones on subagents (FORCE_PROMPT_CACHING_5M, ENABLE_PROMPT_CACHING_1H): over 53 chunks, the worker's 1-hour writes cost $217 at list prices against $233 for 5-minute ones, and subagents' 5-minute writes cost $226 against $289 for 1-hour ones.
  • A security-pass agent type with fewer tools: its 49 runs started at a median 30k tokens and cost $14 in all.
  • Deferred tool loading: already on, since ENABLE_TOOL_SEARCH defaults on and saves MCP tool definitions only.
  • Context editing and programmatic tool calling: API features that Claude Code doesn't expose as a setting.
  • LSP, repo maps and outline scripts: item 231's trial lost findings and agents didn't call them; the only figures found since are a vendor's and a blog's.
  • Compact test reporters and output filters such as rtk: this archive's test output is a few thousand tokens per release, and the filters' figures are their own; trimming read output is item 311's ceiling.

149. Fill each release by priority up to a capacity it states, in place of the bar, and move what does not fit to a later release.

A release's capacity is a sum of its items' cost, as the formula counts it. Each project states its own; the human sets the first value per project. Nothing is deleted by score.

165. Make this repo's tests assert the facts a prompt or help text carries in place of its sentences, and drop the tests that scan source or assert a constant.

A test audit across lilleman's projects on 2026-10-16 found 10 of 33 sampled tests in this repo pinning wording that is not the contract. Examples of these and other weak tests: many tests in bin/loop_claude/test_loop_claude.py pin whole sentences of the question, archive, prompt or help text, up to about 750 characters. test_every_exit_1_stop_has_its_clause_in_help and test_every_failed_judge_run_has_its_clause_in_help parse the source with ast and list each Stop(...) and Failure(...) call by function name. bin/model_routing/test_ladder.py asserts len(LADDER) == 6. Rewrite each to the technical principles' test bullet: assert the facts the text must carry, such as an item ID, a rung or a model argument, and delete a test no real bug could fail. Before and after rewriting a test, run the mutation tool on the code it covers: every mutant the old test killed, the new one kills.

392. Group bin/'s packages by concern into seven folders (plan, records, integrations, loop, review, forge and host), so a newcomer finds a tool's neighbours in one place.

The 0.11.0 release review's architecture pass, round 1, finding 9: bin/ holds 43 packages, about 20 launchers and 10 root files at one level, and the README's two lists are its only index. lilleman chose, during that review, to file this regrouping for a later release instead of doing it in 0.11.0, since it moves every package and no defect lies behind it.

Group bin/ by concern, one package level per group, with the launchers left at the root:

  • plan/: the plan core, with plan_store, release_plan, todo_item and touch_predictions.
  • records/: record_store, project_record and project_config.
  • integrations/: atlassian, adf_codec, rest_client and notifier.
  • loop/: loop_claude, loop_effects, loop_state, model_routing, transcript_archive and routing_figures.
  • review/: larv_review, branch_chunks, review_rules, judge_replay, review_replay, rung_replay, score_outcomes, metric_bench, reading_cost, command_patterns and mutate_changes.
  • forge/: chunk_pr, chunk_preflight and deploy_release, where item 287's shared forge module lands.
  • host/: git_cli, command_lines, claude_code, machine_load, run_heavy, run_tests, text_files, wait_for, rule_hook and subagent_rung.

repo_checks stays at the root, since it tests the repo, not a tool. The README's bin/ list then becomes seven short groups. That list is a place the comprehension seats found hard (README.md → the bin/ package list): make it easier to read.

151. Keep 10 to 20 settled decisions per project as its goals' test set, and rerun them after a goal changes.

Each case is a question the project settled and the answer it chose. A fresh-context agent given only the README's premises answers each case; a case it answers differently says the goals no longer decide it. Written guidance can fail to change behaviour, which is why the goals get a test.

68. Test workflow-bench/compare.py's blinding and bench.py's run plan.

"Most, if not all, code should have automated tests", in principles/technical-principles.md: compare.blind() keeps the comparison blind, and a version name it misses invalidates the bench silently.

The tests also cover workflow-bench/bench.py's permission_flags, with a probe of a headless claude given its flags. It is denied a read outside the granted trees, a bare git status, and wrapper calls with --output, --no-index, a glued -aF/abs, --fil=/abs or a path outside the clone. A cd <elsewhere> && python3 <wrapper> … acts on the clone only. It is allowed the wrapper's log, diff, show, status, ls-files, add and commit -m. workflow-bench/test_review_git.py tests the wrapper's refusals directly.

102. Give loop-claude's retry delay, and the file it reads the machine's memory from, environment overrides, so the black-box test stops replacing time.sleep and machine_load.memory.MEMINFO through sitecustomize.

Found by the principles audit of PR #242. The SITECUSTOMIZE block in bin/loop_claude/test_black_box.py swaps the global time.sleep, the shape "Functional core, imperative shell" bans, and the audit of PR #376 found the same block swapping machine_load.memory.MEMINFO, the constant in bin/machine_load/memory.py. An override of each lets the test run the code as production runs it.

135. Fingerprint only the modules loop-claude imports, so editing another tool's code no longer reloads a running loop.

The maintainability senior and the inherited architect of PR #257's scoring run named fingerprint and exec_if_edited in bin/loop_claude/loop/reload.py: the fingerprint covers every bin/**/*.py, so editing statusline.py or rank_todo.py re-execs a running loop. Make fingerprint easier to read.

5. Key a project on its checkout's full path, and refuse to start while another session uses that key.

The key names the project's session, archive folder and models entry. project_key, in bin/transcript_archive/project_key.py, strips ~/code/ and then any leading /, so ~/code/a and /a share all three, and one loop can resume or delete the other's session.

bin/reading-cost keys the archive by the main checkout, the parent of git's common dir, since it runs in a chunk's worktree; loop-claude keys it by --show-toplevel, so a loop started inside a worktree archives where reading-cost never looks. Derive both keys one way.

291. Let principles/technical-principles.md → "Sane defaults" allow a library whose README goals ask that it work with no config to default the clock and randomness in its public constructor, where a documented option overrides each.

"Sane defaults" has a library ask for every input that can change between runs, the clock and randomness included, and leaves defaulting them to the code that wires a program together. A logging library's principles audit found that its constructor defaults both, as its first goal, useful with no config, asks. Its decision log records the defaults as a departure from the principle, and names this item as what ends it. Until then, each principles audit there may raise the defaults again.

2026-11-10, the human: keep the library's defaults, and give the principle an exception for this case, worded about as the tagline says. The exception has to square with "Functional core, imperative shell", which the same bullet cites and no other principle overrides.

156. Drop rank_todo.py's ranking of an old-shape table once no project under ~/code holds one.

While rank_todo.py ranks two shapes, projects follow two sets of scoring rules, against goal 1.3. Once every project has reshaped, drop OLD_SHAPE's weighing from bin/plan/ranking.py, RESHAPE, and the old shape's lines in rank_todo.py --help and rules/todo-items.md. Keep todo_table reading old tables: score_outcomes reads every past merge's todo.md. PR #283's paired ruling found the old-shape path made ranking.py's score, Scoring.weights, unknown_columns and todo_table.py's Moves harder to read; leave each easier, with one shape and moves that are always moves, and the Moves cell's rules in one place. Release 0.3.0's seats named ranking.py's score, weigh and rank and todo_table.py's OLD_SHAPE for the same fork: "every rule exists twice for the old table shape" (Mid A). Make weigh and rank easier to read too.

388. Refuse an agent's Write or Edit of a routing ladder file in rule-hook, since a change to the ladder is the human's.

Filed by PR #553, which made the routing ladder a file: bin/model_routing/routing-ladder.json, and a project's own routing-ladder.json in its config's folder. CLAUDE.md → "Working loop", "Ask, don't guess" reserves a change to the routing ladder for the human, and until PR #553 the ladder was code, which only a reviewed chunk changed. A worker can now write its project's file, and with it choose the rungs its own later cycles and reviews run on. The security pass of PR #553 noted that a ladder changed mid-cycle also makes rule-hook's reviewer-rung skip its check, since the rungs loop-claude handed are then off the ladder.

Project principle 4 asks that a rule a program can check is enforced by a program. Refuse a Write, an Edit and a MultiEdit of a file named routing-ladder.json under ~/.claude-projects/, with no override, and say that the ladder is the human's to change. The shipped file stays editable in the setup's own checkout, where a chunk's review covers it.

260. Compare the first 20 chunks whose worker got the reading rule with the 20 before them, on review rounds past each pass's first, unfinished verdicts, 🔴, 🟡 and misread findings, dollars and hours per chunk, and drop the rule where any rose.

Start once 20 chunks have run under the rule. The rule is READING_RULE in bin/loop_claude/cycle/prompts.py, which every fresh worker prompt holds; git log --first-parent -S READING_RULE on the default branch finds the merge that added it. A chunk got it when its first worker started after that merge. Measure as docs/decisions.md → loop-claude: the worker's claude compacts its context before it fills a 200k window… measured the window: rounds past each pass's first and the 🔴, 🟡 and misread counts from the round commit titles, unfinished verdicts from the outcomes files, hours from the worker's first message to its last, and dollars from the chunk's cost-state entry. Compare means, since the decision weighs the rule by all spend.

51. Hand loop-claude's terminal output to an injected effect, so no storage or retry function prints.

"Functional core, imperative shell", in principles/technical-principles.md: say in bin/loop_effects/effects.py takes the effects but prints to stderr itself, and retried and the rest print through say. Two bare print(file=sys.stderr) calls remain outside them, in bin/agent_cycle/interrupt.py and bin/loop_claude/project/model_choice.py. The tests swap the global stream.

201. Save the human's tmux setups and restore them after a reboot, so work is up and running again quickly.

The human asked on 2026-10-21 for tmux support, above all for saving tmux setups. Don't build anything until the human has said exactly what they need. Ask them at least:

  • What a setup holds, such as sessions, windows, panes, each pane's working directory and the command running in it, like one loop-claude per project.
  • When a setup is saved: on demand or on its own.
  • How a setup comes back after a reboot.
  • Whether the code that saves and restores setups must work on every machine that takes this checkout (goal 2), which decides whether that code lives in this repo.
  • What else tmux support should cover.

Weigh the widely used tmux-resurrect and tmux-continuum plugins against a script of our own. Then re-score this item from the answers and that choice, and split it into items where it is more than one chunk.

267. Keep waiting when an interrupt gets an empty answer, at the terminal or in the claude opened for it, in place of stopping the loop.

11 runs in the archive ended with "Stopped: the human is needed to answer the question above, but the interrupt ended without anything typed", each leaving its chunk open until the human restarted the loop. That stop comes from human_typed in bin/agent_cycle/interrupt.py, after the claude opened for /talk or for a release ready to cut ends with no human turn. An empty line at the terminal stops the loop the same way, in typed_answer. Item 268 removes one of the cases that opens claude.

4. Name each chunk's archive folder by the chunk's start time, so <key>/ lists chunks in the order they ran.

Today the folder is the SHA-256 of the transcript's first timestamped line, which tells the human nothing; that line carries the start time. Drop the fallback to the first line, which every chunk of a project shares, and give the tests' fake transcripts ISO timestamps. bin/reading-cost takes a chunk's id from the folder's first 12 characters and skips a chunk whose id docs/reading-cost.md holds, so give it a way to match a figure written under the old name.

324. Count in the routing figures the finished chunks whose worker wrote no Chunk items: line.

docs/decisions.md → routing: loop-claude reads the items a chunk takes … holds only while workers write the line when asked, and nothing counts a chunk whose worker skipped it. There loop-claude scores no item beyond the one the chunk was routed on, so the route holds. Count in bin/routing_figures/, beside the chunks counted, those whose worker's transcript holds no such line, so the human sees when the decision's premise stops holding.

187. Count the tokens a worker spends reading reviewers' reports, so goal 1.4's signal shows whether report files lowered what a review costs the worker's context.

Before PR #310, reports reached the worker as the Agent call's result, a median 19k tokens per chunk (2026-10-08, 68 archived transcripts). Now a reviewer returns its verdict and one line per finding, and the worker reads a 🔴 or 🟡 finding's entry from the report file. reading-cost counts only tracked files (bin/reading_cost/measure.py), and routing-figures counts only the Agent call's result (bin/routing_figures/chunk.py), so neither sees the report reads. Found by the architecture review of PR #310.

266. Move the chunks archived under the key home/lilleman/.claude to the key claude-config, so the routing figures and score-outcomes count this repo's chunks as one project.

The archive holds 277 chunk folders under home/lilleman/.claude, from 2026-09-24 on, and 39 under claude-config (2026-11-10). The routing figures of 2026-11-04 list no project home/lilleman/.claude. They find a chunk's later fixes on the default branch of the checkout its key names (bin/routing_figures/archive.py measure_project), and /home/lilleman/.claude is a release checkout on a detached HEAD. Item 5 fixes how keys are derived, not the chunks already archived. ~/.claude/loop-claude-models.json still holds a home/lilleman/.claude entry; it is machine state outside the repo, so removing it is a one-off step on each machine, not part of the change.

181. Have the test of a quoted answer's step-up assert the stepped rung's facts as literals, not through prompts.rungs_held.

bin/loop_claude/test_loop_claude.py's test_a_quoted_answer_that_cites_an_escalation_trigger_steps_the_worker_up_before_it_resumes_and_archives_the_step builds its expected prompt from prompts.rungs_held, the code under test, so a bug there passes it. The technical principles' test bullet bans that shape. Assert the new rung and one review launch as literals, as the typed answer's step-up test does. Raised by the principles audit of PR #308.

193. Narrow numbered's mark to Mark by a typed lookup in place of typing.cast, so a mark added to the regex but not the lookup fails at run time, and one added to the lookup but not Mark fails mypy.

Found by review of PR #317. bin/plan/goals_file.py's numbered writes typing.cast(Mark, mark or None), which asserts the type goal_weights branches on without checking it: a mark added to the regex reaches goal_weights as a value Mark lacks. mypy does not read the regex, so no lookup lets it catch that. Replace the cast with a lookup typed dict[str, Mark], from each mark the regex captures, and "" for none, to its Mark value. A mark added to the regex but not the lookup then raises KeyError at run time, and a value added to the lookup but not Mark fails mypy.

194. Make test_waits_out_a_usage_limit_which_ends_a_run_of_failures expect a usage-limit wait independent of the wall-clock time that passes during the run.

Found by review of PR #317. The black-box test, bin/loop_claude/test_black_box.py:609, not the unit test of the same name in test_loop_claude.py, which runs on a fake clock, sets resets_at from the test's time.time() and expects the wait, rounded to tens, to be 3720. When more than about 5 s pass between that read and the loop's, the wait rounds to 3710 and the test fails with no bug. Item 102 covers only replacing the time.sleep swap; this item changes the expected value, for example by asserting the wait against resets_at + 120 minus the time the loop read, or a bound on it.

265. Archive each wait that costs a chunk wall time, for a usage limit, low memory, a failed worker start or a heavy slot, with its start and end stamps, so goal 1.5's hours split waiting from work.

No record holds these waits today. An outcomes file holds verdicts, stops, step-ups and the judge's cost (bin/transcript_archive/outcomes.py), and ~/.local/state/run-heavy/ holds only the slots' lock files. loop-claude waits out a usage limit and low memory in bin/agent_cycle/retry.py; run-heavy waits for a slot in bin/run_heavy/. A heavy slot's wait happens inside a worker's Bash call, so loop-claude doesn't see it; run-heavy would write it where the archive can find it.

169. Have rule-hook's todo-rank-check catch a todo.md that the same Bash call edits and then commits.

The hook judges a call before it runs, and changed in bin/rule_hook/todo_check.py compares todo.md with the index and HEAD as they stand then. A call that edits todo.md and commits it, such as a heredoc script followed by git commit, passes unchecked: PR #296's commit 149108e did, and the override marker it carried was never logged. larv-review's gate still runs rank_todo.py --check, so an unranked plan stops at merge rather than at commit. A git pre-commit hook would see the staged file.

197. Let the human edit what they type at loop-claude's terminal questions with the arrow keys, so Left moves the cursor in place of printing ^[[D.

bin/loop_effects/effects.py hands loop-claude Python's builtin input, and nothing under bin/ imports readline, so the terminal gets no line editing: fixing a typo means erasing back to it. loop-claude asks two questions at its terminal. Both read through effects.input: the pick of the project's model, in project/model_choice.py's ask_for_model, and the interrupt's answer, in agent_cycle/interrupt.py's typed_answer. Importing readline in effects.py gives input line editing, at two costs. It moves the prompt from stderr to stdout, while the question above it stays on stderr. test_black_box.py waits for the prompt on stderr. readline's bracketed paste takes a pasted block whole, so an answer can hold several lines. worker_prompts.typed_answer ends the answer at its first blank line, so the judge would read the rest of such a paste as loop-claude's words. The product-owner review of PR #308 raised the interrupt's case. lilleman, 2026-10-18: "Not super-important, but a bit irritating."

1. Say in README → Install how the transcript archive is backed up, once bender-infra backs up ai-sandbox's data volume.

It lives under ~/.claude, which .gitignore keeps from the remote, so a lost disk loses every unattended run's record. Back it up only to a destination private to me and encrypted at rest: the worker runs with permissions bypassed, so the archived tool results can hold credentials it read.

2026-10-02, the human: ai-sandbox is the VM loop-claude will run on, hosted on the server bender. bender-infra, the infrastructure-as-code project that manages bender, will back up ai-sandbox's data volume once ai-sandbox is rebuilt. The item waits for the human to say the rebuild has landed; do not remind them. Before writing the README line, check that ~/.claude lives on that volume and that the backup is private and encrypted at rest.

The README line covers the bench's transcripts in workflow-bench/logs/ too (goal 3).

397. Script each step that the chunks of releases 0.10.0 to 0.14.0 repeated by hand and a program can compute, filing one item per step, and file this audit again for the next release.

lilleman asked, on 2026-10-16, for a standing audit each release, so the loop keeps moving work a program does as well from the model to scripts (goals 1.4 and 1.5). This audit reads the archived transcripts of the chunks that shipped releases 0.10.0 to 0.14.0 for every step that chunks repeated by hand, before, during or after review. It keeps the steps whose answer a program computes the same way each time, such as a lookup, a check, a prompt assembled for a reviewer, a cleanup. File one item per such step, with the time and tokens it cost, into the next release with items left. Leave a step that needs judgement with the model. Run command-patterns <version> first for each of 0.10.0 to 0.14.0; each run's table is the audit's starting point. read_threads and calls in bin/command_patterns/tally.py open the calls of one pattern, or command-patterns <version> --calls <file> dumps them all as JSON lines. The 0.7.0 audit found that most steps its chunks repeated were already scripted by that release's own tools, so a step a shipped item already covers is not filed.

A pattern's call count is a poor guide: most of the 0.9.0 calls were reading and hand edits, which stay with the model. Cluster the calls by their command text with numbers and paths normalised away, then check each cluster against the tools in ~/.claude/bin before filing.

The 0.8.0 audit filed four steps, items 334, 335, 336 and 337. The 0.9.0 audit filed one, item 396, and found most other repeated steps already covered by tools that shipped inside the release: chunk-pr merge's wait for the checks, run-tests, wait-for, larv-review round-title and todo-item show.

The chunk that ships this item files a copy of this audit in the release after this item's own, auditing the releases cut since, and marks the copy owner, so the audit recurs each release until lilleman drops it.