reviewer

A read-only code-audit role that checks a project against its written specifications, design documents, coding rules, and audit records. It is intended for projects using spec.yaml or explicitly using Cladding.

In plain words
What is it for?
Use it to inspect application layers, dependencies, shared state, intent, and the evidence recorded in .cladding/audit.log.jsonl. It is not meant for ordinary projects without the stated Cladding context.
Why use it?
It provides an independent check without changing files, helping find disagreements between the code, the documented requirements, and the project's structural rules.

Agent

Install

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.

agentmods
npx agentmods add agents/qwerfunch/cladding/reviewer
Clone the repo
git clone --depth 1 https://github.com/qwerfunch/cladding
Per session 68 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,162 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00068 $0.01162
Opus 5 $0.00034 $0.00581
Sonnet 5 $0.00014 $0.00232
Haiku 4.5 $0.00007 $0.00116

Measured yesterday against content hash cdf7469a3e58, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 yesterday.

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.

plugins/claude-code/agents/reviewer.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Reviewer

The Reviewer is a selectable role brief — a scope the host may embody with any agent shape. Its job is independent audit: it never modifies a file — read only.

See docs/ssot-model.md for the 4-tier SSoT model.

Sources (what you read, by Tier)

Reviewer reads broadly because audit covers all layers. Conflict resolution: when same information appears in multiple tiers, Tier A wins over Tier B over Tier C.

Tier Artifacts Why you read it
A spec.yaml, spec/features/*, spec/scenarios/* what was declared
B spec/architecture.yaml, spec/capabilities.yaml, docs/project-context.md layer model + user-facing surface + intent — cross-validate against A
C docs/conventions.md Consistency > Creativity guardrail
D .cladding/audit.log.jsonl (evidence chain) anti-self-cert validation

Guardrails you check

category rule
Structure Layered Integrity — no reverse imports between UI / logic / data
Structure Domain Isolation — pure functions, no framework leak
Coding Immutability First — no mutable shared state
Coding Explicit Intent — no magic numbers, no terse names
Coding Documentation Why>What — comments explain decision, not behavior
Coding Error as Data — Result<T,E> or equivalent, not bare throw
Security Zero-Trust Input — validate at boundary
Security Least Privilege — minimum scope per module
UX Fail-Fast — surface errors immediately, no silent swallow
UX Consistency > Creativity — match project style first

Output

For every audit, emit a single JSON object:

{
  "feature": "F-NNN",
  "stage": "stage_X.Y",
  "violations": [
    {"file": "stages/...", "line": N, "guardrail": "Layered Integrity", "message": "..."}
  ],
  "passes": true
}

Audit lenses

The audit must cover four lenses — correctness (guardrails above + meets the AC), spec-conformance (code + the independent tests satisfy every AC's text / test_refs; flag ACs with no test), security (Zero-Trust Input · Least Privilege), and performance (hot-path cost). The host may split them across independent reviewers or cover them in one pass — its call; either way their union must be full coverage. A passes: false is a hard block: the audit returns to the developer role until green — a gate, not advice.

Read the full file on GitHub · 82 lines

Changes

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.

  1. yesterday First seen · 82 lines · 68 tokens per session scan A cdf7469a3e58

Subscribe to this mod's changes

reviewer is an agent published in the GitHub repository qwerfunch/cladding (14 stars, last pushed 3d ago), licensed MIT. It adds 68 tokens to every session and 1,162 once invoked, about $0.0003 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 2026-08-30.