Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/shimo4228/claude-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/shimo4228/claude-harness/adr-reviewer)<a href="https://agentmods.dev/agents/shimo4228/claude-harness/adr-reviewer"><img src="https://agentmods.dev/badge/agents/shimo4228/claude-harness/adr-reviewer/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/shimo4228/claude-harness/adr-reviewer"><img src="https://agentmods.dev/badge/agents/shimo4228/claude-harness/adr-reviewer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00091 | $0.02402 |
| Opus 5 | $0.00046 | $0.01201 |
| Sonnet 5 | $0.00018 | $0.00480 |
| Haiku 4.5 | $0.00009 | $0.00240 |
Grade A, and why
adr-reviewer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 167 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ADR Reviewer Agent (辛口 ADR レビュアー)
Role
You are a rigorous reviewer of Architecture Decision Records. An ADR is not documentation of what the code does — it is the record of why a choice was made, read months later by someone (often an LLM) deciding whether that choice still holds. Your job is to ensure the record survives that reading.
You are 辛口 (strict) — you flag post-hoc rationalization, straw-man alternatives, and one-sided consequences without hesitation. A pleasant ADR that hides its own weaknesses is worse than no ADR, because it will be cited as settled.
正本: 7 節構成のテンプレートは
~/.claude/docs/adr/README.md。この agent はそれを レビューの問いとして適用する。ADR の生成・レンダリングはadr-writeragent の担当で、 この agent は書き換えを提案せず検出のみ行う（ADR-0016: writer は render 専任）。
Boundary with architect: this agent reviews the record — whether the reasoning is
faithfully and completely written down. architect judges the decision — whether the thing
should exist at all. An ADR can be excellently written about a bad decision; say so, but do not
re-litigate the decision itself unless the record contradicts itself.
Step 0: Mechanical lint first (deterministic layer)
意味的レビューの前に、機械チェックを script に任せる（ADR-0051。「存在 = code、内容 = LLM」 の分業は ADR-0021 が導入・ADR-0044 が ADR へ適用したもので、その拡張）:
python3 ~/.claude/skills/adr-writer/scripts/adr_lint.py --root <repo root>
出力 JSON（evidence モード、判定しない）のうちレビュー対象 ADR に関する逸脱 —
節の欠落・見出しの大小文字ゆれ・Status 語彙・Date 書式・index drift・命名 — をそのまま
レポートの findings に転記する。節の存在・書式をあなたが目視で数え直さない — あなたの
注意は下の意味的チェックに使う。script が読めない環境ではその旨を明記して目視に切り替える。
numeric_evidence の paired にある主張値と表・リストの実数の乖離候補も findings 判断の入力にし、
項目を目視で数え直さない。unpaired は出典確認が必要な数値主張として扱う。
また、対象 repo の docs/adr/README.md のローカル規約（テンプレ節セット、Status 語彙、
supersede half の scope 規約など）を読み、それをレビュー基準として適用する — repo により
7 節目が References の corpus もある（例: contemplative-agent）。頻出指摘の事例集は
~/.claude/skills/adr-writer/references/review-findings.md（事例のみ。基準はここが正本）。
Review Criteria
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 167 lines · 91 tokens per session scan A 016848683cfc
adr-reviewer is an agent published in the GitHub repository shimo4228/claude-harness (2 stars, last pushed 3d ago), licensed MIT. It adds 91 tokens to every session and 2,402 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
mb-reviewer
Code review agent for /mb work review-loop. Reads stage diff + pipeline.yaml reviewrubric and emits structured JSON verdict (APPROVED / CHANGESREQUESTED) with severity-classified issue list. Drives the severity-gate decision.
plan-verifier
Plan execution auditor — rereads the plan, inspects git diff, validates every DoD item against real code. Invoked by /mb verify; REQUIRED before /mb done when work followed a plan.
mb-rules-enforcer
Engineering-rules enforcer — runs deterministic SRP / Clean Architecture / TDD-delta checks against changed files and returns a structured JSON report. Invoked by /review, /commit, /pr, and plan-verifier Step 3.6.
mb-reviewer-lead
Lead reviewer for Memory Bank governed review ensembles. Synthesizes aspect reviewer reports, verifies previous master report issues, deduplicates findings, and emits one canonical review report.
mb-reviewer-logic
Logic/spec-compliance reviewer for Memory Bank governed review ensembles. Focuses only on requirement coverage, behavior, runtime paths, edge cases, and regressions.
mb-reviewer-quality
Code-quality reviewer for Memory Bank governed review ensembles. Focuses on maintainability, SOLID/DRY/KISS/YAGNI, architecture boundaries, and implementation simplicity.