About Argus
Argus is an MCP server published by pixelesq in the Developer Tools category: reverse-engineering evidence from a running Windows process, not the binary on disk. It has been installed 0 times through Conduid.
The repository has 6 stars and 0 forks, with the last commit 6 months ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues (0) before depending on it in production.
Install
npx argusclaude mcp add argus -- npx -y @itonskie/argusnpx -y @itonskie/argusuvx argus-testingpip install argus-testingThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Ask AI
Ask AI about Argus
Powered by Claude · Grounded in docs
Security checks
- ·README presentNot checked yet.
- ·License declaredNot checked yet.
- ·Tests presentNot checked yet.
- ·Dependencies pinnedNot checked yet.
- ·No dynamic code executionNot checked yet.
- !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.
Releases
README
The Problem
Enterprise SEO tools cost $100-300/month. SEO expertise is locked behind years of specialist knowledge. Most free tools only scratch the surface.
Argus breaks these barriers. It's a completely free, open-source Chrome extension that gives you instant, deep SEO analysis on any webpage — powered by Claude Opus 4.6 for AI insights that rival expensive platforms.
Features
Three-Tab Architecture
Extract — Every SEO tag at a glance
- Title, meta description, canonical, robots (with character count badges)
- Open Graph & Twitter Card tags with live social preview cards
- JSON-LD structured data with syntax highlighting and validation
- Heading hierarchy as a visual tree
- Link analysis (internal/external/nofollow counts)
- Image audit (missing alt text, dimensions, lazy loading, modern formats)
- One-click copy and full JSON/text export
Audit — 40+ rule-based SEO checks
- Weighted scoring (0-100) across 10 categories
- Categories: title, description, headings, images, links, technical, structured data, social, content, performance
- Color-coded severity (critical/warning/info/pass) with fix guidance
- Real Web Vitals: LCP, INP, CLS, TTFB via PerformanceObserver API
Insights — AI-powered deep analysis
- Three-tier AI provider system with automatic fallback
- Automatic page classification, content quality, and SEO strategy analysis
- On-demand: rewrite meta descriptions, improve titles, suggest FAQ schema, generate schema markup
- Opus 4.6 exclusive: SEO strategy brief, technical fixes, content gap analysis, competitor insights
Argus uses Claude Opus 4.6 as its primary AI engine via a BYOK (Bring Your Own Key) architecture:
Three-Tier AI Provider System
| Tier | Provider | How It Works |
|---|---|---|
| 1 (Primary) | Claude Opus 4.6 | User provides their API key (stored locally, never synced). Direct fetch() from the side panel to api.anthropic.com with streaming SSE. |
| 2 (Fallback) | Gemini Nano | Chrome's on-device AI model. No API key, no network — runs locally. |
| 3 (Always Available) | Static Analysis | Rule-based analysis — works offline with zero dependencies. |
Opus 4.6 Exclusive Features
When Claude is configured, Argus unlocks advanced analysis that smaller models can't handle:
- SEO Strategy Brief — Comprehensive prioritized action plan analyzing ALL extracted data (meta, headings, links, images, schema, technical signals) plus audit results
- Schema Markup Generator — Generates complete, valid JSON-LD ready to paste into your page
- Technical Fix Generator — Produces exact HTML code fixes for every audit issue
- Content Gap Analysis — Identifies missing sections based on page type and heading structure
- Competitor Insights — E-E-A-T signals, featured snippet optimization, and advanced tactics
Why BYOK?
- Privacy — Your API key never leaves your browser (stored in
chrome.storage.local) - No backend — Direct browser-to-API calls, no proxy server
- No cost to us — Users control their own usage
- Graceful degradation — Works perfectly without any API key (Gemini Nano or static analysis)
API Architecture
Side Panel (React)
│
├─ Has Claude key? → fetch('https://api.anthropic.com/v1/messages')
│ with streaming SSE, max_tokens: 2048
│
├─ Gemini Nano available? → Chrome Built-in AI Prompt API (on-device)
│
└─ Neither? → Static rule-based analysis (always works)
Argus includes a Model Context Protocol (MCP) server that brings SEO auditing directly into Claude Code. Developers can audit any webpage's SEO without leaving their terminal.
Tools
| Tool | Description |
|---|---|
seo_audit |
Full 40+ rule audit with weighted scoring |
extract_meta |
Extract all meta tags, OG, Twitter, JSON-LD, headings, links, images |
compare_seo |
Side-by-side comparison of 2-5 URLs |
extract_json |
Raw extraction + audit data as JSON for programmatic use |
Setup
One-line install (recommended):
claude mcp add argus-seo -- npx -y @pixelesq/argus-mcp
Or build from source:
# Build the MCP server
cd mcp-server && npm install && npm run build
# Add to Claude Code
claude mcp add argus-seo -- node mcp-server/build/index.js
Or drop a .mcp.json in any project:
{
"mcpServers": {
"argus-seo": {
"command": "node",
"args": ["path/to/argus/mcp-server/build/index.js"]
}
}
}
Shared Audit Engine
The MCP server shares the same audit engine as the Chrome extension. The audit rules are pure functions that take a PageExtraction data structure — they work identically in both browser and Node.js runtimes.
Chrome Extension MCP Server
───────────── ──────────
content.ts (DOM APIs) fetch() + cheerio (HTML parsing)
↓ ↓
PageExtraction ◄───── SAME TYPE ─────► PageExtraction
↓ ↓
auditors/engine.ts ◄── SHARED CODE ──► auditors/engine.ts
↓ ↓
AuditReport AuditReport (returned via MCP)
Chrome Extension
Quick install: Add to Chrome from the Chrome Web Store.
Or build from source:
- Clone and build:
git clone https://github.com/pixelesq/argus.git cd argus npm install npm run build - Open
chrome://extensions/, enable Developer mode - Click Load unpacked, select
.output/chrome-mv3 - Click the Argus icon — the side panel opens
Configure AI (Optional)
Claude Opus 4.6: Click the gear icon in the Insights tab → enter your API key from console.anthropic.com
Gemini Nano: Requires Chrome 138+. Enable chrome://flags/#prompt-api-for-gemini-nano, restart Chrome.
Tech Stack
| Layer | Technology |
|---|---|
| Extension Framework | WXT (Vite-based) |
| UI | React 19 + TailwindCSS 4 |
| Language | TypeScript (strict) |
| Manifest | Chrome Extension Manifest V3 |
| AI (Cloud) | Claude Opus 4.6 via Messages API |
| AI (On-Device) | Chrome Built-in AI (Gemini Nano) |
| MCP Server | @modelcontextprotocol/sdk + cheerio |
| Icons | Lucide React |
Project Structure
argus/
├── entrypoints/ # Chrome extension entry points
│ ├── background.ts # Service worker (side panel + headers)
│ ├── content.ts # Content script (DOM extraction)
│ └── sidepanel/ # React side panel UI
│ ├── App.tsx
│ ├── tabs/ # Extract, Audit, Insights
│ └── components/ # UI components
├── lib/ # Core business logic
│ ├── ai/ # AI provider system
│ │ ├── claude.ts # Claude Opus 4.6 API client
│ │ ├── gemini.ts # Gemini Nano client
│ │ ├── provider.ts # Unified provider with fallback
│ │ ├── prompts.ts # System prompts (standard + Opus-enhanced)
│ │ └── settings.ts # API key management
│ ├── auditors/ # Audit engine (shared with MCP)
│ │ ├── engine.ts # Orchestration
│ │ ├── scoring.ts # Weighted scoring algorithm
│ │ └── rules/ # 40+ audit rules across 10 categories
│ └── extractors/ # DOM-based data extraction
├── mcp-server/ # MCP server for Claude Code
│ ├── src/
│ │ ├── index.ts # Server + tools
│ │ └── extractor.ts # cheerio-based HTML extraction
│ └── package.json
└── .mcp.json # MCP config for Claude Code
Privacy
- No analytics, no tracking, no data collection
- The extension makes zero network requests except:
- (Optional) HEAD request for HTTP response headers
- (Optional) Claude API calls — only when user provides their own key, direct to Anthropic
- Gemini Nano runs entirely on-device
- API keys stored in
chrome.storage.local(device-local, never synced)
Development
# Extension dev server (hot reload)
npm run dev
# Production build
npm run build
# MCP server
cd mcp-server && npm install && npm run build
License
MIT License — see LICENSE for details.
README mirrored from the source repository 19 days ago. The original is authoritative.