Skip to main content
Glama
AIEthos-ai

mcp-ethostron-live

Official
README.md
# EthosTron™ Live MCP Server ⚡

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Model Context Protocol](https://img.shields.io/badge/MCP-Protocol%20Compliant-green)](https://modelcontextprotocol.io/)
[![Vitest](https://img.shields.io/badge/Tests-3%2F3%20Passing-brightgreen)](https://vitest.dev/)
[![Status](https://img.shields.io/badge/Cloud%20Run-Live-emerald)](https://aiethos.ai)

The official **Model Context Protocol (MCP)** server for [AIEthos](https://aiethos.ai).

**EthosTron™ Live** brings real-time Generative Engine Optimization (GEO) directly into your AI coding assistant (Claude Desktop, Cursor, Antigravity, Windsurf, VS Code). In under 2 seconds, your agent can audit any website for AI search readiness, inspect crawler access (`GPTBot`, `PerplexityBot`, `ClaudeBot`, `Google-Extended`), detect missing Schema.org entities, and synthesize drop-in **Inverted-Pyramid Direct-Answer blocks** engineered for citation across frontier models.

---

## ⚡ Multi-Assistant Setup & Configuration

EthosTron™ Live supports both local MCP (stdio JSON-RPC) for IDE coding assistants and standard OpenAPI Actions for cloud LLM assistants like ChatGPT:

### 1. GitHub Copilot (Visual Studio Code)
GitHub Copilot natively supports MCP servers directly within VS Code.

Add to your project's `.vscode/mcp.json` (or run Command Palette `Ctrl+Shift+P` / `Cmd+Shift+P` > **Copilot: Manage MCP Servers**):

```json
{
  "mcp": {
    "servers": {
      "ethostron": {
        "command": "npx",
        "args": ["-y", "@aiethos/ethostron-mcp"]
      }
    }
  }
}
```
**Copilot Chat Usage:**
```text
@ethostron Audit https://example.com for GEO and identify missing schema types
```

---

### 2. ChatGPT (OpenAI Custom GPTs & Actions)
For OpenAI ChatGPT (Plus, Team, Enterprise), connect EthosTron Live via **OpenAPI Actions** pointing to the Cloud Run backend without needing a local daemon:

1. Open ChatGPT > **Explore GPTs** > **Create** (or edit an existing Custom GPT).
2. Go to the **Configure** tab > Scroll down to **Actions** > Click **Create new action**.
3. Under **Schema**, click **Import from URL** and enter:
   ```text
   https://aiethos-backend-978593064341.us-west2.run.app/openapi.json
   ```
4. **Authentication:** Leave as `None` (the public developer `/public/mcp/*` endpoints are open & read-only).
5. **Privacy Policy:** `https://aiethos.ai/privacy`
6. Save and publish your Custom GPT!

**ChatGPT Usage:**
```text
Audit https://stripe.com for GEO and generate a direct answer remediation block.
```

---

### 3. Google Gemini & Antigravity
For developers building with Google Gemini models or Google Antigravity:

**Option A — Antigravity & Gemini CLI (`.gemini/config/mcp_config.json`):**
```json
{
  "mcpServers": {
    "ethostron": {
      "command": "npx",
      "args": ["-y", "@aiethos/ethostron-mcp"]
    }
  }
}
```

**Option B — Google GenAI Python SDK (`google-genai`):**
```python
from google import genai
from google.genai import types

client = genai.Client()
# Connect to EthosTron Live OpenAPI endpoints directly
response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Audit https://example.com using aiethos_geo_quickscan"
)
```

---

### 4. Claude Desktop
Add to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "ethostron": {
      "command": "npx",
      "args": ["-y", "@aiethos/ethostron-mcp"]
    }
  }
}
```

*Config file locations:*
- **macOS:** `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows:** `%APPDATA%\Claude\claude_desktop_config.json`
- **Linux:** `~/.config/Claude/claude_desktop_config.json`

**Claude Usage:** Click the tool hammer icon in Claude Desktop and prompt: *"Audit https://linear.app for GEO"*.

---

### 5. Cursor & Windsurf
In **Settings > Features > MCP Servers**, add a new server:
- **Name:** `ethostron`
- **Type:** `command`
- **Command:** `npx -y @aiethos/ethostron-mcp`

Or add directly to `.cursor/mcp.json` or `.windsurf/mcp.json`:
```json
{
  "mcpServers": {
    "ethostron": {
      "command": "npx",
      "args": ["-y", "@aiethos/ethostron-mcp"]
    }
  }
}
```

---

### 6. VS Code Autonomous Agents (Cline / Roo Code / Continue)
Add to your extension's MCP configuration (e.g. `~/.cline/mcp_settings.json` or Roo Code MCP settings):

```json
{
  "mcpServers": {
    "ethostron": {
      "command": "npx",
      "args": ["-y", "@aiethos/ethostron-mcp"]
    }
  }
}
```
Autonomous coding agents will automatically invoke `aiethos_geo_quickscan` to diagnose missing Schema.org markup or robots directives, and call `aiethos_preview_ethostron_fix` to apply verified drop-in code patches.

---

## 🛠️ Available MCP Tools

### 1. `aiethos_geo_quickscan`
Performs an instant, sub-2.0s Generative Engine Optimization assessment for any URL:
- **AI Crawler Permissions:** Checks `robots.txt` directives for `GPTBot` (OpenAI), `PerplexityBot`, `ClaudeBot`, and `Google-Extended` (Gemini).
- **Schema.org Microdata Health:** Inspects JSON-LD for `Organization`, `TechArticle`, `FAQPage`, `SoftwareApplication`, and `Product`.
- **Inverted-Pyramid Readability:** Measures opening lead paragraph word count against the optimal LLM answer extraction target (45–65 words).
- **Deterministic 0–100 Score:** Provides immediate numerical grading and high-impact critical gaps.
- **Claim Link:** Includes an activation URL to upgrade to full multi-page audits.

```text
// Example Agent Prompt:
"Audit https://stripe.com for Generative Engine Optimization using aiethos_geo_quickscan"
```

**Example Response Output:**
```json
{
  "domain": "stripe.com",
  "url": "https://stripe.com",
  "geoScore": 85,
  "crawlerStatus": {
    "gptBot": "allowed",
    "perplexityBot": "allowed",
    "claudeBot": "allowed",
    "googleExtended": "allowed"
  },
  "schemaHealth": {
    "hasJsonLd": true,
    "detectedTypes": ["Organization", "WebSite"],
    "missingRecommendedTypes": ["FAQPage", "TechArticle"]
  },
  "directAnswerReadability": {
    "invertedPyramidDetected": true,
    "openingParagraphWordCount": 54,
    "status": "optimal",
    "recommendation": "Opening paragraph is concise and optimized for LLM answer extraction."
  },
  "criticalGaps": [
    "Missing high-citation schema (FAQPage or TechArticle recommended)"
  ],
  "claimUrl": "https://aiethos.ai/activate?domain=stripe.com&claim_token=mcp_8f2a1b9c&utm_source=ethostron_mcp&utm_medium=ai_copilot&utm_campaign=geo_quickscan"
}
```

---

### 2. `aiethos_preview_ethostron_fix`
The automated content remediation engine: Synthesizes a copy-paste ready, 45–60 word Direct-Answer block and matching Schema.org microdata tailored to your topic and intent category.

Supported Intent Topologies:
- `definitive_architecture`: High-authority entity definitions (what it is, core mechanism, business impact).
- `comparative_benchmark`: Objective side-by-side technical trade-offs and decision criteria.
- `troubleshooting_resolution`: Exact error signatures, root cause explanations, and CLI commands.
- `pricing_commercial`: Transparent tier tables, inclusions, and ROI formulas.
- `framework_methodology`: Sequential numbered workflows with CLI verification steps.

```text
// Example Agent Prompt:
"Use aiethos_preview_ethostron_fix to generate an authoritative direct answer block for 'SOC2 Compliance & Zero Trust' on https://acme.io with intent_category 'definitive_architecture'"
```

---

### 3. `aiethos_check_llm_citations`
Checks public prompt benchmarks to determine whether a brand is actively cited across frontier LLMs (OpenAI, Gemini, Claude) for common industry prompts, or whether competitors dominate citations.

```text
// Example Agent Prompt:
"Check AI citation share for Datadog in Cloud Observability using aiethos_check_llm_citations"
```

---

## 📋 Available Prompt Templates

- **`audit-website-geo`**: Complete workflow guiding your AI assistant through scanning, diagnosing, and delivering remediation advice for any target URL.
- **`remediate-citation-gap`**: Turns any identified content gap into a publish-ready Inverted-Pyramid markdown block and JSON-LD schema snippet.

---

## 🎯 How the MCP GEO Score is Calculated

The MCP GEO score is a **deterministic 100-point formula** evaluating the 3 technical pillars of AI search citation:

| Pillar | Max Points | Measurement Criteria |
| :--- | :---: | :--- |
| **1. AI Crawler Access** | **30 Pts** | Inspects `robots.txt` for `GPTBot` (+10), `PerplexityBot` (+10), `ClaudeBot` (+5), and `Google-Extended` (+5). |
| **2. Schema.org Health** | **40 Pts** | Valid JSON-LD present (+15), `Organization`/Brand entity grounded (+10), High-citation schema present (+15). |
| **3. Direct-Answer Readability** | **30 Pts** | Inverted-pyramid opening paragraph: **45–65 words** (+30). Suboptimal lengths (+15). Missing (+5). |

---

## 🔄 Upgrading to Full Multi-Page Audits

The MCP Quickscan is an instant single-page check. To unlock **continuous brand governance**, click the `claimUrl` returned in the tool response:

- **Full Domain Crawl:** Deep 20-page crawl with client-side SPA JavaScript rendering.
- **Web Verification:** Tavily search queries indexing 3rd-party coverage and reviews.
- **Multi-Model Consensus:** Parallel evaluation across OpenAI, Gemini, and Claude with hallucination risk penalties.
- **Autonomous Remediation:** Automated GitHub Pull Requests that apply verified JSON-LD patches.

---

## 🛡️ Privacy, Security & Quotas

- **Strictly Read-Only:** EthosTron Live never mutates local files, executes git commits, or modifies production databases.
- **SSRF Protection:** Target addresses are validated against private network CIDRs (`127.0.0.1`, `10.0.0.0/8`, `192.168.0.0/16`, `169.254.169.254`).
- **Free Anonymous Tier:** 5 quickscans per domain per day with zero setup or API key.
- **Zero Database Pollution:** Anonymous scans do not create dummy accounts or pollute customer dashboards.

---

## 💻 Local Development & Testing

```bash
# Clone repository
git clone https://github.com/AIEthos-ai/ethostron-mcp.git
cd ethostron-mcp

# Install dependencies
pnpm install

# Build TypeScript
pnpm run build

# Run unit tests
pnpm test

# Test JSON-RPC via MCP Inspector
npx @modelcontextprotocol/inspector node dist/index.js
```

---

## 📄 License

Apache-2.0 © [AIEthos LLC](https://aiethos.ai)

TDQS

B3.2/5.0

Scored across 3 tools

Disambiguation4/5

Each tool maps to a distinct stage: technical GEO audit (quickscan), sample content synthesis (preview_ethostron_fix), and live citation tracking (check_llm_citations). The two diagnostic tools (quickscan vs. citation check) could be momentarily confused since both assess AI visibility, but the descriptions clearly separate on-site technical readiness from off-site citation presence.

Naming Consistency3/5

All names share the aiethos_ prefix and snake_case, which is good, but the verb conventions are mixed: check_llm_citations is verb-first while geo_quickscan and preview_ethostron_fix are noun/adjective-first. Readable, but not a predictable pattern an agent can generalize from.

Tool Count3/5

Three tools is thin for a GEO platform surface; the set reads as a minimal entry tier. Each tool does earn its place, but the server feels borderline under-scoped rather than well-rounded.

Completeness3/5

The lifecycle stops at diagnosis and a sample: quickscan audits, preview_ethostron_fix only produces a 'taste test' block, and nothing generates the full fix, applies schema markup, exports reports, or tracks changes over time. Agents hitting the preview tool have no path to the actual remediation output, a notable dead end.

Maintenance

ActivityMaintained
ResponsivenessNo issues