reviewer-architecture

reviewer-architecture is an agent for Claude Code from wsauret/flywheel. It costs 239 tokens per session (1,469 once invoked), scanned A, original, MIT.

A code-review agent that examines the architecture of software changes, including dependency direction, layer boundaries, state ownership, and system design patterns.

In plain words
What is it for?
Reviewing pull requests and feature changes for architectural fit, checking how modules depend on one another, and assessing whether new behavior respects the project's existing design.
Why use it?
It can reveal circular dependencies or changes that place logic in the wrong part of an application, even when the code itself works.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter.

Part of the flywheel plugin — 15 skills, 15 agents shipped together

Good fit Reviewing pull requests and feature changes for architectural fit, checking how modules depend on one another, and assessing whether new behavior respects the project's existing design.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/wsauret/flywheel/reviewer-architecture
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.

Clone the repo
git clone --depth 1 https://github.com/wsauret/flywheel

Made for: Claude Code.

Or install flywheel, the plugin that ships this one along with the rest of its 15 skills, 15 agents.

Wrote 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.

agentmods badge for reviewer-architecture

README.md
[![agentmods](https://agentmods.dev/badge/agents/wsauret/flywheel/reviewer-architecture.svg)](https://agentmods.dev/agents/wsauret/flywheel/reviewer-architecture)
Your own site
<a href="https://agentmods.dev/agents/wsauret/flywheel/reviewer-architecture"><img src="https://agentmods.dev/badge/agents/wsauret/flywheel/reviewer-architecture.svg" alt="Measured on agentmods" height="20"></a>
Per session 239 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,469 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00239 $0.01469
Opus 5 $0.00120 $0.00734
Sonnet 5 $0.00048 $0.00294
Haiku 4.5 $0.00024 $0.00147

Measured 8d ago against content hash 973f77e2841d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

reviewer-architecture 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 8d 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.

flywheel/agents/reviewer-architecture.md · 89 lines

How it starts

The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are an architecture reviewer who traces dependency direction, state ownership, and abstraction layer boundaries. You spot layering violations and circular imports by mentally mapping the module graph.

You are NOT checking whether the code is correct, well-typed, or consistent with naming conventions — other reviewers handle that. You are checking whether the change respects the system's boundaries, dependency direction, state ownership model, and abstraction layers.

Phase 0: Load Project Context

The orchestrator passes project context paths in the dispatch under "PROJECT CONTEXT PATHS." Read those paths to learn the project's layer boundaries, state ownership, event patterns, interface conventions, and lifecycle rules. Do not search for additional docs — the orchestrator already discovered them.

Identify the tech stack (state management, event system, DI patterns) from the paths and the code under review.

If the dispatch says "PROJECT CONTEXT PATHS: none," apply the universal principles below.

Review Process

1. Layer Boundaries and Dependency Direction

  • Do imports flow in the correct direction? Are there new circular dependencies?
  • Does the change keep responsibilities within the right module/layer?
  • Does it reach into another component's internals?

2. State Ownership

  • Single source of truth: Is each piece of state owned by exactly one layer/module? Flag changes that create parallel state — the same value stored in two places, or intermediate layers that copy/transform/relay state.
  • State mechanism selection: If the project defines which mechanism to use for which purpose (e.g., stores for UI-read state, event bus for infrastructure notifications, callbacks for 1:1 lifecycle signals), verify the change uses the right one.
  • Derived vs. stored: Is the change storing something that could be derived from existing state?

3. Event-Driven Architecture

  • Producer/consumer coupling: Do event producers remain unaware of their consumers? Flag producers that filter or transform events for specific consumers — subscribers should own their filtering.
  • Mechanism appropriateness: If the project defines different mechanisms for different consumer patterns (broadcast vs. 1:1, state vs. notification), verify the change uses the correct one.
  • No cascading mutations: Flag event handlers that trigger other event handlers in a chain, creating implicit ordering dependencies.

Read the full file on GitHub · 89 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. 8d ago First seen · 89 lines · 239 tokens per session scan A 973f77e2841d

Subscribe to this mod's changes

reviewer-architecture is an agent published in the GitHub repository wsauret/flywheel (14 stars, last pushed 4d ago), licensed MIT. It adds 239 tokens to every session and 1,469 once invoked, about $0.0012 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.

Related

Other agents, from other repositories

code-reviewer

Code reviewer. Delegate only when the user explicitly starts an Octopus workflow.

nyldn/claude-octopus · 19 tokens

edge-case-explorer

Systematically discovers and catalogs edge cases that should be covered by tests for a given piece of code. Traces input sources, call chains, and integration boundaries to find boundary values, type coercion traps, external input messiness, state-dependent failures, and error propagation gaps. Use when exploring how…

testdouble/han · 135 tokens

code-reviewer

Review code changes against a base branch with structured feedback. Use this agent when the user requests a code review, PR review, or wants to analyze code changes systematically.

NikiforovAll/claude-code-rules · 37 tokens

adversarial-validator

Assumes investigation evidence is WRONG and the proposed fix will FAIL. Searches for counter-evidence, unhandled edge cases, and flawed assumptions. Use for adversarial validation of investigation findings and planned fixes.

testdouble/han · 45 tokens

reviewer

Philosophical guardrails enforcer — independently audits code, tests, and spec for layered-integrity, Why>What, error-as-data, and the related Ironclad philosophical invariants. Activate only when the connected project contains spec.yaml or the user explicitly names Cladding; ignore ordinary requests in uninitialized…

qwerfunch/cladding · 68 tokens

ia-best-practices-researcher

Researches external framework docs, version-specific constraints, and industry conventions for any technology. Use when you need authoritative external documentation.

iliaal/whetstone · 34 tokens