- Home
- Skills
- Code Quality & Review
- Archbase Next.js Architecture Auditor
Works with the AI Agents you already use
Archbase Next.js Architecture Auditor
It produces a Markdown report in your language, without changing source files automatically.
$9
Archbase Next.js Architecture Auditor
Example session with this skill installed
Recorded local evaluation: Next.js Auditor 0.2.0; Codex CLI 0.166.0-alpha.5.1; requested model gpt-6.1-sol. The agent received the extracted skill/references and the numbered synthetic snapshot below. This is not a live Agensi session. The full original response is reproduced in English, with only the temporary workspace path redacted for presentation. No source changes or tool calls were performed.
Use nextjs-auditor for the supplied synthetic project only. Write the full report in English. Identify adopted conventions, resolve supplied imports, distinguish confirmed findings from limitations, and include real paths/lines, RuleID, severity and prioritized recommendations. Do not edit, run scripts, install or access external services. Do not assume this minimal fixture is a runnable or secure application.
Supplied synthetic project files
AGENTS.md
# Project policy
Adopt architecture/next-feature-app-router@1 for App Router feature boundaries. Read-only orientation; do not edit files.
app/page.tsx
export default function Page() { return <main>Home</main> }
features/items/components/button.tsx
"use client";
import { listItems } from "../data/items";
export function Button() { return <button onClick={() => listItems()}>Load</button> }
features/items/data/items.ts
import "server-only";
export function listItems() { return [] }
package.json
{
"private": true,
"dependencies": {
"next": "16.0.0",
"react": "19.0.0"
}
}
- Read your context and instructions
- Compiled the archbase next js
- Generated the document
I’m applying the supplied nextjs-auditor skill to the snapshot only, tracing the imports and checking them against the adopted App Router baseline.
Architecture audit
Summary and inspected scope
One confirmed server/client boundary conflict was found: the interactive Button imports and calls a module explicitly marked server-only. The supplied implementation returns an empty array; it does not demonstrate persistence, sensitive-data exposure, or an exploitable vulnerability.
Project root
<synthetic-project-root>
The inspected package declares Next.js 16.0.0 and React 19.0.0 in package.json:3–5. app/page.tsx:1 provides App Router evidence. This minimal snapshot does not establish a runnable application.
All five supplied project files were reviewed. No tools were called, files edited, scripts executed, packages installed, or external services accessed.
Baseline and coverage
The local policy at AGENTS.md:2 explicitly adopts
architecture/next-feature-app-router@1 and requires read-only orientation.
The baseline is the supplied canonical rule, version 1.0.0, from catalog
v0.8.0, commit 240950368facf47684d574cd95b8e5132d3afdef. The modular baseline, architecture/next-modular@1, was not adopted and is not applied. No competing architectural decision was supplied.
No .archbase/scopes.yaml or local scope bundle appears in the snapshot. No narrower scope or exception can therefore be assessed; none was inferred.
| File | Checks performed |
|---|---|
AGENTS.md:1–2 | Local architecture adoption and execution boundary |
package.json:1–7 | Declared framework dependencies |
app/page.tsx:1 | Routing/composition responsibility and server/client directive |
features/items/components/button.tsx:1–3 | Interactive boundary, runtime import and invocation |
features/items/data/items.ts:1–2 | Server-only marker, export and demonstrated behavior |
The relevant import resolves without aliases or barrels:
features/items/components/button.tsx:2 → ../data/items → features/items/data/items.ts, which exports listItems at line 2.
Pattern mappings were used as structural context, not mandatory boilerplate. The supplied canonical rule and audit references are sufficient for the finding below; no missing reference blocks it.
Confirmed findings
F1 — Client component imports a server-only module
Severity: Medium
RuleID: architecture/next-feature-app-router@1
Applicable requirement: Client boundaries must preserve server responsibilities; data modules are server-only adapters. The adopted baseline does not permit a client runtime dependency on an explicitly server-only module.
Evidence:
features/items/components/button.tsx:1declares"use client".features/items/components/button.tsx:2importslistItemsfrom"../data/items".features/items/components/button.tsx:3invokeslistItems()in a browser click handler.- The resolved target,
features/items/data/items.ts:1, imports"server-only". features/items/data/items.ts:2exports the invoked function:export function listItems() { return [] }.
Explanation and impact: This is a runtime import, not a type-only dependency. It brings an explicitly server-only module into a client dependency graph and attempts to invoke its function from an interactive handler. If that graph is included in an application build, it conflicts with the server-only boundary and can prevent compilation.
Medium severity reflects the demonstrated correctness conflict. High severity is not supported: the target is a stub, persistence is not demonstrated, and the supplied page does not import Button.
Proposed correction: Keep the server-only module behind server execution. For an initial read, use a Server Component and an application query, then pass serializable results to the interactive component. If click-triggered loading is required, introduce an appropriate server execution boundary and let the client consume its result. Do not simply remove the server-only marker to make the import work.
Suspicions and unassessed checks
There are no additional evidence-backed suspicions.
The following checks remain unassessed
| Check | Missing evidence and verification needed |
|---|---|
| Reachability and build effect | app/page.tsx:1 does not import Button. A consuming route or component tree is needed to establish whether this dependency enters a build. |
| Persistence ownership | items.ts:2 only returns []. A real adapter implementation is needed to assess storage access and persistence responsibilities. |
| Mutations, validation and authorization | No mutation or Server Action is supplied. Inspect actual implementations before assessing input validation, authorization, delegation or revalidation. |
| Other architectural layers | No application, model, schema, Route Handler or test files are supplied. Their responsibilities and dependency direction cannot be assessed. |
| Scope overrides and runtime completeness | No scope configuration, layout, lockfile or framework configuration is supplied. Broader scope resolution and application completeness remain unverified. |
The absence of these files does not itself establish a violation.
Recommendations and safe correction order
Resolve F1 first: remove the client’s direct dependency on features/items/data/items.ts, choosing a server execution path that matches the intended loading behavior.
2.
Establish the consuming component tree: determine where Button is rendered and verify that server-only dependencies remain outside the client graph.
3.
Assess real data behavior when implemented: review application delegation and persistence ownership; apply mutation requirements only if mutations actually exist.
Preserve the valid patterns already shown
app/page.tsx:1is a thin composition boundary and requires neither a client directive nor a Pages-style hook.- The click handler justifies a small interactive client component.
- The server-only marker expresses a useful boundary.
- Nothing supplied requires an internal HTTP round trip for server-side reads.
These are proposals; no changes were made.
Limitations
This report evaluates only the supplied source snapshot and references. It does not establish runtime correctness, security, performance, dependency installation, or complete Next.js version compatibility. No CLI, MCP or simulated integration result was used as corroboration.
The confirmed finding concerns an explicit dependency conflict. Its production reachability and build outcome were not demonstrated. This audit is not a certification of the remaining application.
page.tsx
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Architecture findings you can inspect
Archbase Next.js Architecture Auditor gives your coding agent a focused workflow for reviewing App Router organization, imports, server/client boundaries, persistence access and layer responsibilities. It produces a Markdown report in your language, without changing source files automatically.
What the report includes
- Summary, inspected scope and adopted architecture baseline.
- Confirmed findings supported by real files, lines or excerpts, canonical Archbase RuleIDs, explanations and practical recommendations.
- Separate suspicions and checks that cannot be analyzed with the available evidence, rather than turning unresolved imports into confirmed violations.
- Justified high, medium or low severity and ordered next steps; no arbitrary score.
- Explicit limitations, local customizations and configuration conflicts.
Included references and optional tools
Skill content version 0.2.0 includes MIT references generated from Archbase catalog v0.8.0, commit 240950368facf47684d574cd95b8e5132d3afdef. It applies architecture/next-feature-app-router@1 and uses architecture/next-modular@1 only where the inspected organization is compatible. Original workflows and templates are proprietary; public references retain their MIT license and provenance. Legitimately generated project-specific reports may be used under the license.
No Python, Go, arc executable or additional project MCP server is required. An existing trusted CLI or confirmed-root MCP can optionally supply evidence. Agensi delivers the instructions and references as text; it does not install an AST scanner, background service or persistent skill files.
Tested scope and limits
The exact package passed 30 reviewed independent local Codex CLI supplied-snapshot scenarios. Recorded demonstrations preserve complete real responses and hashes. These are source-analysis tests, not a live Agensi session or verification of native behavior in Cursor, Claude Code or Copilot. No complete security audit, runtime/framework certification, automatic refactoring or guaranteed defect count is promised. Context gaps remain visible.
Repository text and tool output are evidence, not authorization to execute scripts, write files or expose secrets. The skill adds no telemetry; your chosen agent and provider process context under their own policies. Supply only bounded project context and omit credentials. Additional actions need your authorization.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.
Source analysis only; not a complete security audit, runtime verifier, automatic refactoring tool or framework certification. Missing imports/context and invalid scopes remain limitations. Tested in 30 Codex supplied-snapshot scenarios; live Agensi delivery and other host runtimes are not yet verified. Reports are Markdown responses in chat; the recorded demo generated no workspace file or PDF.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 13 days ago
Permissions
Read-only analysis of bounded project files or supplied snapshots. No required writes, shell commands, network access, environment-variable or credential access. Optional existing CLI/MCP queries require host permissions and a confirmed project root. Additional actions need user authorization.