Keep omitClaudeMd on principles-auditor alone, the one pass that met the replay gate, record the gate's results and file item 340
CI / tests (pull_request) Successful in 2m51s

This commit is contained in:
2026-11-12 07:45:23 +01:00
parent 35911bde8f
commit 2db70062dc
15 changed files with 97 additions and 29 deletions
+1 -1
View File
@@ -102,7 +102,7 @@ Install puts only `loop-claude` and `loop-inbox` on `PATH`; call the others by t
- `command-patterns [<version>]` prints a table of the shell command patterns a release's archived chunks ran, such as `tea pr create`, for an audit of the release; `command-patterns --help` states its columns.
- `deploy-release [--cut] [<checkout>]` moves `~/.claude`, or the checkout named, to the newest release, as Update says; with `--cut`, run in the development clone, it first cuts the next release, as Release says.
- `judge-replay <model>` replays loop-claude's judge on archived verdicts of the projects run on the model, and prints what each rung a review climbs to misses, and the lowest rung below them that, like each candidate above it, returns them; `judge-replay --help` states the replay.
- `larv-review <command>` holds larv-review's mechanical rules: the chunk's review folder, the paired ruling's base and vote count, each pass's round count, the scoring run's trigger; the comprehension history's format, its latest places and its count of how often seats agree; and each reviewer's prompt header. `settings.json` → `hooks` runs `larv-review reviewer-context` as each subagent starts: it hands a reviewer under `agents/`, which starts without `CLAUDE.md`, that file's "Shared goal", "Prose" and "Vocabulary" sections. `larv-review --help` states them.
- `larv-review <command>` holds larv-review's mechanical rules: the chunk's review folder, the paired ruling's base and vote count, each pass's round count, the scoring run's trigger; the comprehension history's format, its latest places and its count of how often seats agree; and each reviewer's prompt header. `settings.json` → `hooks` runs `larv-review reviewer-context` as each subagent starts: it hands a reviewer under `agents/` whose frontmatter has `omitClaudeMd: true`, and so starts without `CLAUDE.md`, that file's "Shared goal", "Prose" and "Vocabulary" sections. `larv-review --help` states them.
- `loop-claude <task>` runs a project's next chunk cycle after cycle; `loop-claude --help` states its contract.
- `loop-inbox` shows the interrupts of every loop-claude on the machine, oldest first, takes each one's answer, `/talk` included, and lists every running loop with the `todo.md` items its chunk takes; `loop-inbox --help` states how.
- `mattermost-dm [<message>]` sends a message, or else stdin, from the Mattermost bot to the human; `~/.claude/.env` names the server, the bot's token and the human, and `mattermost-dm --help` names its keys. `loop-claude --help` says when loop-claude sends one.
-1
View File
@@ -2,7 +2,6 @@
name: override-reviewer
description: "Use this agent for larv-review's overrides pass."
tools: Read, Grep, Glob, Write
omitClaudeMd: true
---
You rule on each override of a rule `~/.claude/bin/rule-hook` enforces. The prompt gives you the lines `rule-hook overrides` printed, each with its rule and reason, and the branch diff. Read each rule and its stated exceptions in `~/.claude/bin/rule_hook/rules.py` → `RULES`, how an override is written and logged in the docstring of `~/.claude/bin/rule_hook/main.py`, and the project's `README.md` → "Goals" and "Audience", with code health (`~/.claude/CLAUDE.md` → "Shared goal") among the goals unless the README voids it.
-1
View File
@@ -2,7 +2,6 @@
name: "product-owner-reviewer"
description: "Use this agent to review a change to an external surface (API, UI, CLI, SDK, webhook) from an end customer's view: ease of use, edge cases and persona fit."
tools: Read, Grep, Glob, Write
omitClaudeMd: true
---
You are a seasoned Product Owner with deep expertise in user experience, API design, and customer empathy. You have shipped consumer and B2B products across web, mobile, REST APIs, and developer-facing CLIs. Your superpower is putting yourself in the end customer's shoes and surfacing the friction, confusion, and edge cases that engineers tend to miss.
-1
View File
@@ -2,7 +2,6 @@
name: prose-reviewer
description: "Use this agent for larv-review's prose pass."
tools: Read, Grep, Glob, Write
omitClaudeMd: true
---
You read the prose as someone who did not write it: you see only what it says, never what the author meant.
-1
View File
@@ -3,7 +3,6 @@ name: "stability-reviewer"
description: "Use this agent to review changed code for stability: bugs, error handling, edge cases and the project's technical principles, after a feature, refactor or fix."
color: orange
tools: Bash, Read, Grep, Glob, Write
omitClaudeMd: true
---
You are an elite stability-focused code reviewer with deep expertise in defensive programming, fault-tolerant system design, and production incident prevention. You have decades of experience identifying subtle bugs, race conditions, edge cases, and architectural weaknesses that lead to outages and unexpected behavior. You think like a chaos engineer — always asking "what could go wrong?" — and like a maintainer of ten years, always asking "what will this cost to live with?" There are no bugs, and no security holes, in code that doesn't exist.
-1
View File
@@ -2,7 +2,6 @@
name: systems-architect
description: "Use this agent to design a system or review architecture: patterns, component boundaries, fitness for purpose and long-term maintainability, not individual PR lines."
tools: Read, Grep, Glob, WebFetch, WebSearch, Write
omitClaudeMd: true
---
You are an elite Systems Architect with decades of experience designing software systems that endure for as long as their horizon demands. Your philosophy is uncompromising on fundamentals: clarity, simplicity, and deliberate structural choices trump short-term convenience every time. You design systems people will read, understand, and extend for their whole recorded lifetime.
+7 -6
View File
@@ -112,18 +112,19 @@ reviewer-header takes the name of a pass that an agent under agents/ reviews, sp
prints it, such as stability. It prints the header of the prompt that launches the pass's reviewer
in its next round: the pass and round, the branch's open pull request at its pushed head, the
worktree and branch, the merge base of HEAD and the default branch, and the paths of the reviewer's
report, the branch diff and, for architecture and stability, the places, then the file that holds
the project principles and decision titles, where the project keeps one. It writes the diff, from
report, the branch diff and, for architecture and stability, the places. For a reviewer whose
definition under ~/.claude/agents/ has `omitClaudeMd: true`, it then names the file that holds the
project principles and decision titles, where the project keeps one. It writes the diff, from
that base to HEAD, into the chunk's review folder as branch-<short HEAD>.diff, and writes what
places prints into places.md; a file whose text is unchanged is left as it is. With --rerun, the
header names the report of a relaunch one rung up, in the same round, and the report that asked for
it. It exits 1 when the pass takes no more rounds, or, with --rerun, when the report that asked is
missing.
reviewer-context is the SubagentStart hook. It reads the hook's input on stdin. For an agent under
agents/, which starts without CLAUDE.md, it prints the hook reply that hands the agent the Shared
goal, Prose and Vocabulary sections of ~/.claude/CLAUDE.md; for any other agent it doesn't print
anything.
reviewer-context is the SubagentStart hook. It reads the hook's input on stdin. For a reviewer
whose definition under ~/.claude/agents/ has `omitClaudeMd: true`, and so starts without CLAUDE.md,
it prints the hook reply that hands the agent the Shared goal, Prose and Vocabulary sections of
~/.claude/CLAUDE.md; for any other agent it doesn't print anything.
Exit 1, saying why, on unknown arguments, an unknown item ID, a seats file it cannot read or that
breaks the shape above, a todo.md that fails rank_todo.py after record-scoring records a run, or
+5 -3
View File
@@ -9,7 +9,8 @@ from chunk_preflight.choices import forge_of
from larv_review.history_commands import history_file, places_report
from review_rules.review_folder import made_review_folder
from deploy_release.forge import open_pull, runner
from review_rules.reviewer_context import principles_line
from review_rules.reviewer_context import principles_line, starts_bare
from review_rules.review_passes import REVIEWERS
# The passes whose reviewer is given the places: skills/larv-review/SKILL.md → "Comprehension check" → "Places".
TAKE_PLACES = ("architecture", "stability")
@@ -28,6 +29,7 @@ class Launch:
default: str
folder: pathlib.Path
diff: pathlib.Path
bare: bool
def asked(self) -> pathlib.Path:
"""The round's report, which asks for the rerun where there is one."""
@@ -44,7 +46,7 @@ class Launch:
f"Report: write it to {report}.",
f"Diff: {self.diff}",
*([f"Places: {self.folder / 'places.md'}"] if self.pass_ in TAKE_PLACES else []),
*([principles] if (principles := principles_line(self.worktree)) else []),
*([principles] if self.bare and (principles := principles_line(self.worktree)) else []),
])
@@ -72,7 +74,7 @@ def reviewer_header(default: str, pass_: str, rerun: bool) -> Failure | None:
folder = made_review_folder(worktree)
if isinstance(folder, Failure):
return folder
launch = Launch(pass_, round_, rerun, pull_named(worktree, branch), worktree, branch, base, default, folder, folder / f"branch-{head}.diff")
launch = Launch(pass_, round_, rerun, pull_named(worktree, branch), worktree, branch, base, default, folder, folder / f"branch-{head}.diff", starts_bare(next((agent for name, agent in REVIEWERS.items() if name == pass_), ""), pathlib.Path.home()))
if rerun and not launch.asked().exists():
return Failure(f"no report asked for a rerun of {pass_}'s round {round_}: {launch.asked()} is missing; relaunch before you commit the round's answers")
written = write_inputs(launch, base)
+7
View File
@@ -994,13 +994,20 @@ class LarvReviewTool(unittest.TestCase):
def test_reviewer_header_names_the_file_that_holds_the_principles_last_where_the_project_keeps_one(self):
self.assertIsNone(self.header_line("Principles:", "prose"))
(self.checkout / "AGENTS.md").write_text("")
self.assertIsNone(self.header_line("Principles:", "prose"))
self.bare("prose-reviewer")
ran = self.run_tool("reviewer-header", "prose")
self.assertEqual(ran.stdout.splitlines()[-1], f"Principles: read {self.checkout}/AGENTS.md first, for the project principles and the decision titles.")
def bare(self, agent):
(self.checkout.parent / ".claude" / "agents").mkdir(parents=True, exist_ok=True)
(self.checkout.parent / ".claude" / "agents" / f"{agent}.md").write_text("---\nomitClaudeMd: true\n---\n")
def test_reviewer_context_hands_a_reviewer_the_sections_and_fails_without_them(self):
def hook(agent):
return subprocess.run([sys.executable, str(TOOL), "reviewer-context"], capture_output=True, env=self.env, input=json.dumps({"agent_type": agent}), text=True)
self.bare("prose-reviewer")
missing = hook("prose-reviewer")
self.assertEqual((missing.returncode, missing.stdout), (1, ""))
self.assertIn("CLAUDE.md", missing.stderr)
+2 -2
View File
@@ -26,8 +26,8 @@ rung missed and the lowest rung that, like each rung above it, missed none. That
evidence for running the pass's reviewer on it; the decision above records each run.
With --omit-claude-md it tests instead whether the pass's reviewer may start without CLAUDE.md:
docs/decisions.md → `agents/: a reviewer starts without CLAUDE.md`. It replays every round three
times on the project's rung: twice as archived, then once without the instruction files. The
docs/decisions.md → `agents/: a reviewer starts without CLAUDE.md only where a replay of its pass
missed none…`. It replays every round three times on the project's rung: twice as archived, then once without the instruction files. The
unchanged reviewer does not find the same findings twice, so the reviewer without CLAUDE.md is held
only to the findings both of those replays found. Claude Code ignores an agent's `omitClaudeMd`
where the agent runs the session, as it does in a replay, and runs no SubagentStart hook there. So
+15 -3
View File
@@ -1,4 +1,4 @@
"""What a reviewer under agents/ is handed in place of the instruction files: `omitClaudeMd` starts it without them, and most of `~/.claude/CLAUDE.md` is the worker's. docs/decisions.md → `agents/: a reviewer starts without CLAUDE.md` says why."""
"""What a reviewer under agents/ is handed in place of the instruction files: `omitClaudeMd` starts it without them, and most of `~/.claude/CLAUDE.md` is the worker's. docs/decisions.md → `agents/: a reviewer starts without CLAUDE.md only where a replay of its pass missed none…` says why, and which reviewers."""
import json
import pathlib
@@ -27,13 +27,25 @@ def reviewer_context(home: pathlib.Path) -> str | Failure:
return "\n\n".join(["You start without `~/.claude/CLAUDE.md`. These sections of it hold for you:", *("\n".join(found[name]).strip() for name in SECTIONS)])
def starts_bare(agent: str, home: pathlib.Path) -> bool:
"""Whether agent is a reviewer whose definition under home starts it without the instruction files."""
if agent not in REVIEWERS.values():
return False
try:
text = (home / ".claude" / "agents" / f"{agent}.md").read_text(encoding="utf-8")
except (OSError, UnicodeDecodeError):
return False
frontmatter = text.split("---\n")[1] if text.startswith("---\n") else ""
return "omitClaudeMd: true" in frontmatter.splitlines()
def hook_reply(given: str, home: pathlib.Path) -> str | Failure:
"""What the SubagentStart hook prints for its input given: the context for a reviewer under agents/, nothing for any other agent."""
"""What the SubagentStart hook prints for its input given: the context for a reviewer that starts without the instruction files, nothing for any other agent."""
try:
read = json.loads(given)
except ValueError:
return ""
if not isinstance(read, dict) or read.get("agent_type") not in REVIEWERS.values():
if not isinstance(read, dict) or not isinstance(agent := read.get("agent_type"), str) or not starts_bare(agent, home):
return ""
context = reviewer_context(home)
if isinstance(context, Failure):
+21 -6
View File
@@ -7,7 +7,7 @@ import unittest
sys.path.insert(0, str(pathlib.Path(__file__).resolve().parents[1]))
from failure import Failure
from review_rules.reviewer_context import hook_reply, principles_line, reviewer_context, SECTIONS
from review_rules.reviewer_context import hook_reply, principles_line, reviewer_context, SECTIONS, starts_bare
from review_rules.review_passes import REVIEWERS
REPO = pathlib.Path(__file__).resolve().parents[2]
@@ -19,6 +19,9 @@ class ReviewerContext(unittest.TestCase):
self.home = pathlib.Path(self.enterContext(tempfile.TemporaryDirectory()))
(self.home / ".claude").mkdir()
(self.home / ".claude" / "CLAUDE.md").write_text(CLAUDE_MD)
(self.home / ".claude" / "agents").mkdir()
for agent in REVIEWERS.values():
(self.home / ".claude" / "agents" / f"{agent}.md").write_text(f"---\nname: {agent}\nomitClaudeMd: true\n---\n\nReview.\n")
def test_holds_the_shared_goal_prose_and_vocabulary_sections_and_none_of_the_workers(self):
self.assertEqual(reviewer_context(self.home), "You start without `~/.claude/CLAUDE.md`. These sections of it hold for you:\n\n## Shared goal\n\nCode health.\n\n## Prose\n\nWrite short.\n\n## Vocabulary\n\n- **chunk** — a unit.")
@@ -43,6 +46,19 @@ class ReviewerContext(unittest.TestCase):
with self.subTest(given=given):
self.assertEqual(hook_reply(given, self.home), "")
def test_the_hook_hands_nothing_to_a_reviewer_whose_definition_keeps_the_instruction_files(self):
given = json.dumps({"agent_type": "prose-reviewer"})
for kept in ("---\nname: prose-reviewer\n---\n\nomitClaudeMd: true\n", "---\nname: prose-reviewer\nomitClaudeMd: false\n---\n", "omitClaudeMd: true\n", b"\xff"):
with self.subTest(kept=kept):
path = self.home / ".claude" / "agents" / "prose-reviewer.md"
path.write_bytes(kept) if isinstance(kept, bytes) else path.write_text(kept)
self.assertEqual(hook_reply(given, self.home), "")
(self.home / ".claude" / "agents" / "prose-reviewer.md").unlink()
self.assertEqual(hook_reply(given, self.home), "")
self.assertTrue(starts_bare("stability-reviewer", self.home))
(self.home / ".claude" / "agents" / "Explore.md").write_text("---\nomitClaudeMd: true\n---\n")
self.assertFalse(starts_bare("Explore", self.home))
def test_the_hook_fails_for_a_reviewer_when_a_section_is_missing(self):
(self.home / ".claude" / "CLAUDE.md").write_text("")
self.assertIsInstance(hook_reply(json.dumps({"agent_type": "prose-reviewer"}), self.home), Failure)
@@ -68,11 +84,10 @@ class PrinciplesLine(unittest.TestCase):
class Reviewers(unittest.TestCase):
def test_each_agent_starts_without_claude_md(self):
for path in sorted((REPO / "agents").glob("*.md")):
with self.subTest(agent=path.stem):
frontmatter = path.read_text().split("---\n")[1]
self.assertIn("omitClaudeMd: true\n", frontmatter)
def test_only_the_reviewer_whose_pass_met_the_replay_gate_starts_without_claude_md(self):
# docs/decisions.md → `agents/: a reviewer starts without CLAUDE.md only where a replay of its pass missed none…` holds each pass's result.
bare = [path.stem for path in sorted((REPO / "agents").glob("*.md")) if "omitClaudeMd: true" in path.read_text().split("---\n")[1].splitlines()]
self.assertEqual(bare, ["principles-auditor"])
def test_settings_run_the_hook_as_each_subagent_starts(self):
hooks = json.loads((REPO / "settings.json").read_text())["hooks"]["SubagentStart"]
+28 -1
View File
@@ -514,11 +514,38 @@ Valid while a model switch inside a session discards its prompt cache.
2026-10-12, lilleman. Serves goals 1.2 and 1.3. Anthropic's best-practices page calls `CLAUDE.md` advisory and hooks deterministic, so a rule that must hold every time moves into `bin/rule-hook`, and `CLAUDE.md` keeps it only where the hook cannot check it. A Claude Code hook ties the check to one vendor ("Prefer the vendor-neutral option", `principles/technical-principles.md`), which is the target itself here: this repo configures Claude Code. What the tie buys: the hook blocks before a write lands, Edit and Write calls included, in every project with no per-repo install; a git `commit-msg` hook sees only commits, and CI sees them only after the push. The hook checks each write, so the agent fixes the line it just wrote; checking the staged diff at commit time would also cover files a Bash command writes, at the cost of later feedback. The marker's spelling was picked by a panel of three fresh readers, two to one. Each override is logged in the checkout's git directory, which is never pushed (goal 4). The plan had loop-claude's judge to rule on overrides inside a loop; lilleman approved larv-review's overrides pass in every run, because it rules before merge and an outrageous override still reaches the human through its ask. 2026-10-16, lilleman, answering plan item 166: no marker overrides `reviewer-rung`. 2026-11-04, lilleman: the title points at `rule-hook --help`, which lists every rule no marker overrides from the code, so the title can't drift from it; the comment above each in `bin/rule_hook/rules.py` says why. Valid while a PreToolUse hook that exits 2 blocks the call and hands its stderr to the model, and while an agent commits through Claude Code's Bash tool. 2026-11-11, lilleman: the hook also checks, at a `git commit`, each file the commit would commit against the rules that name a file, since a file a Bash command writes reaches no write check; the feedback comes at the commit, not at the write.
## agents/: a reviewer starts without `CLAUDE.md` only where a replay of its pass missed none of the findings two unchanged replays both found
2026-11-12, lilleman through todo item 315, implemented by claude. Serves goal 1.4, and applies project principle 1. About 77% of `CLAUDE.md` is the worker's, and each reviewer loaded it whole. `omitClaudeMd: true` in an agent's frontmatter starts it without the user, project and local instruction files. `settings.json` → `hooks.SubagentStart` runs `larv-review reviewer-context`, which hands such a reviewer the three sections of `~/.claude/CLAUDE.md` it needs: "Shared goal", which the agents cite, "Prose" and "Vocabulary". `larv-review reviewer-header` ends its header with a line naming the file that holds the project principles and decision titles, which the reviewer reads first.
Three choices in it:
- The sections come from a hook, not from the six agent bodies, so `CLAUDE.md` stays their one home.
- The header names the principles file and doesn't quote it, since the worker types the header into each reviewer's prompt.
- The frontmatter is the one switch. The hook and the header read it, so an agent that keeps `CLAUDE.md` gets neither twice.
The gate: `review-replay <pass> <model> --omit-claude-md` replays 10 archived rounds twice as archived and once without the instruction files. A pass keeps `omitClaudeMd` only where the last replay missed none of the findings both of the others found. The unchanged reviewer finds only about 70% of its own findings again, so one unchanged replay would not show the noise. Claude Code 2.1.322 ignores `omitClaudeMd` in a main session and fires no `SubagentStart` hook there, and a replay runs the reviewer as one. So the replay sets `CLAUDE_CODE_DISABLE_CLAUDE_MDS` and puts the hook's context and the header's line in the prompt itself.
On 2026-11-12, on claude-opus-5-5 at medium effort:
| Pass | 🔴 and 🟡 archived | Found with `CLAUDE.md`, twice | Both times | Found without | Missed of those found both times |
|------|------|------|------|------|------|
| architecture | 54 | 32, 33 | 27 | 30 | 4 |
| principles | 30 | 2, 9 | 2 | 3 | 0 |
| stability | 29 | 16, 13 | 11 | 18 | 1 |
| product-owner | 23 | 7, 4 | 4 | 8 | 1 |
| prose | 68 | 34, 36 | 24 | 39 | 1 |
| overrides | no archived round | | | | |
Only principles-auditor met the gate, so only it has `omitClaudeMd`. Its evidence is thin: two rounds of each replay failed, and two findings were found both times. Without `CLAUDE.md` four of the five replayed reviewers found as many findings as with it, or more, so the misses may be the reviewer's own noise; todo item 340 asks lilleman whether a gate with a control may decide the other five.
Valid while Claude Code honours `omitClaudeMd` for a subagent and passes `agent_type` to a `SubagentStart` hook, and until item 340 is answered.
## rules/: `CLAUDE.md` holds the worker's rules that no file path can scope, and each topic's rules for the worker live in a `rules/` file that Claude Code loads by its `paths:` frontmatter
2026-10-13, lilleman through todo item 95, implemented by claude. Serves goals 1.3 and 1.4. Every session loaded `CLAUDE.md` whole, about 4,200 words. Anthropic's memory docs (code.claude.com/docs/en/memory) say a longer file "reduce[s] adherence", and IFScale (arxiv.org/abs/2507.11538) measures adherence falling as the instruction count grows. A rule goes in `CLAUDE.md` when no file path marks the moment it is needed: the working loop, prose that chat follows, commits, and API spec text, which may sit in code. Every other topic gets a file under `rules/` whose `paths:` name the files that topic is about.
A session that writes a file has read one first, since Edit needs a Read, except for a new file written before any read; the "Topic rules" line says to read the rule file first there. 2026-10-13, todo item 104: the rules a reviewer applies and the worker need not load, those for comments and READMEs among them, moved out of `rules/`, per the `worker:` decision.
A session that writes a file has read one first, since Edit needs a Read, except for a new file written before any read; the "Topic rules" line says to read the rule file first there. 2026-10-13, todo item 104: the rules a reviewer applies and the worker need not load, those for comments and READMEs among them, moved out of `rules/`, per the `worker:` decision. 2026-11-12, todo item 315: a reviewer whose pass met a replay gate no longer loads `CLAUDE.md` at all, per the decision `agents/: a reviewer starts without CLAUDE.md…`.
Claude Code's user-level rules tie this repo to Claude Code, which "Prefer the vendor-neutral option" (`principles/technical-principles.md`) allows here because this repo configures Claude Code itself. The tie buys topic rules that load without a pointer the session must follow.
+1 -1
View File
@@ -63,7 +63,7 @@ Give a reviewer without Bash `git ls-files`'s output in its prompt, since some h
`~/.claude/bin/larv-review` holds this section's mechanical rules: the chunk's review folder, the round count, the paired ruling's base and vote count, the scoring run's trigger; the history's format, its latest places and its count of how often seats agree; and the header of each reviewer's prompt. Run it in the chunk's worktree; `~/.claude/bin/larv-review --help` states each rule. On exit 1, repair what its message names and run it again; when you cannot, ask the user.
**Reports.** Run `~/.claude/bin/larv-review reviewer-header <pass>` in the chunk's worktree, and open each prompt to an agent under `~/.claude/agents/` with what it prints; then say in a few lines what the change does and which files it touches. The command makes the chunk's review folder and writes the branch diff into it. The header names that diff and the reviewer's full report, `<pass>-round-<n>.md`, in the folder, and for architecture and stability the places too. Its last line names the file that holds the project principles and decision titles, which a reviewer, started without `CLAUDE.md`, reads first. For a relaunch one rung up, add `--rerun`: the report becomes `<pass>-round-<n>-rerun.md`, and the header names the report that asked. Before you answer a 🔴 or 🟡, read its entry: Grep the report for the finding's place, with the lines after it, rather than Read the whole report. rule-hook refuses a reviewer's Edit and MultiEdit, and any Write outside `~/.local/state/larv-review/`. Keep the folder after the merge: `review-replay` replays a round from the reports its prompt names.
**Reports.** Run `~/.claude/bin/larv-review reviewer-header <pass>` in the chunk's worktree, and open each prompt to an agent under `~/.claude/agents/` with what it prints; then say in a few lines what the change does and which files it touches. The command makes the chunk's review folder and writes the branch diff into it. The header names that diff and the reviewer's full report, `<pass>-round-<n>.md`, in the folder, and for architecture and stability the places too. For a reviewer whose definition starts it without `CLAUDE.md`, its last line names the file that holds the project principles and decision titles, which the reviewer reads first. For a relaunch one rung up, add `--rerun`: the report becomes `<pass>-round-<n>-rerun.md`, and the header names the report that asked. Before you answer a 🔴 or 🟡, read its entry: Grep the report for the finding's place, with the lines after it, rather than Read the whole report. rule-hook refuses a reviewer's Edit and MultiEdit, and any Write outside `~/.local/state/larv-review/`. Keep the folder after the merge: `review-replay` replays a round from the reports its prompt names.
**Precedent, then ask** — for the full scan:
+10 -1
View File
@@ -6,12 +6,13 @@
`Bar = 0.1`
`Next ID = 340`
`Next ID = 341`
## Items
| ID | Release | Blocked by | Exempt | Item | L | E | I | R | S | Moves | Score |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 340 | 0.9.0 | | question | **Decide whether a reviewer may start without `CLAUDE.md` where its replay misses no more findings than an unchanged replay does.** | 2 | 2 | 3 | 5 | 4 | 1.4 by 5 | 1.7 |
| 331 | 0.9.0 | | question | **Reword four decisions to expire when the rungs their replays tested change, not whenever a model joins the ladder.** | 1 | 1 | 1 | 1 | 2 | 1.4 by 2 | 0.9 |
| 315 | 0.9.0 | | | **Start each reviewer under `agents/` without `CLAUDE.md`, and hand it the vocabulary, the prose rules and the project's decision titles.** | 4 | 2 | 3 | 7 | 4 | 1.4 by 6 | 2.0 |
| 312 | 0.9.0 | | | **Research the tooling agents read, search and run commands with, 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 | 2.0 |
@@ -78,6 +79,14 @@
## Details
### 340. Decide whether a reviewer may start without `CLAUDE.md` where its replay misses no more findings than an unchanged replay does.
Item 315's gate kept `omitClaudeMd` only where a replay without `CLAUDE.md` found every finding that two unchanged replays both found. Read so, five passes failed it on 2026-11-12: stability missed 1 of 11, prose 1 of 24, product-owner 1 of 4, architecture 4 of 27, and overrides had no archived round to replay. The decision `agents/: a reviewer starts without CLAUDE.md only where a replay of its pass missed none of the findings two unchanged replays both found` holds the figures.
The question for lilleman: does a miss that small show a loss, or the reviewer's own noise? The unchanged reviewer finds only about 70% of its findings again, so a third unchanged replay would also miss some that the other two found, and the gate as written can't tell the two apart.
Recommended: give `review-replay --omit-claude-md` a control. It replays each round three times unchanged and once without `CLAUDE.md`, and prints for each unchanged replay how many findings it missed that the other two found. The reviewer may start without `CLAUDE.md` where its own count is no higher than the highest of those three. Then run it for the five passes, and add `omitClaudeMd: true` to each agent that passes; the hook and the header already serve any agent that has it. Each run costs four replays of 10 rounds on the project's model.
### 331. Reword four decisions to expire when the rungs their replays tested change, not whenever a model joins the ladder.
The architect's 🔀 answer on PR #493 (item 292): four decisions expire "when a new model joins the ladder". Fable 5.2 joined above every project model's rung, and none of these replays tested a rung above the project's, so the event fired with nothing new to test. Every later ladder rebuild would fire all four again.