LLM SEO MCP — Elephant Accountability

Elephant Accountability MCP Server

MCP server for the Elephant Accountability certification bureau. Surfaces vendor offerings, methodology, and audit requests to AI agents over the Model Context Protocol.

Live endpoint: https://elephant-mcp.fly.dev/mcp Manifest: https://elephant-mcp.fly.dev/.well-known/mcp.json A2A Agent Card: https://elephant-mcp.fly.dev/.well-known/agent.json Homepage: https://eaccountability.org

What this server does

When an AI agent (Claude, ChatGPT, a custom LangChain agent, etc.) needs to evaluate Elephant Accountability against the EVI v0.9 standard or surface its audit tiers to a buyer, it queries this server instead of scraping a website.

Six tools are exposed:

Two resources are exposed via resources/list: elephant://offerings, elephant://proof-points, elephant://transparency.

Why you'd use this

• You're building a procurement agent and want to query Elephant Accountability's audit tiers and EVI v0.9 methodology directly.
• You're at Claude Desktop / Cursor / any MCP-compatible client and want direct access to Elephant's offerings + fit assessment.
• You're a competitor studying how to deploy your own MCP server — this repo is MIT-licensed, clone freely.

Quickstart — local development

git clone https://github.com/Chris-Eaccountability/elephant-accountability-mcp.gitcd elephant-accountability-mcp
python -m venv .venv && source .venv/bin/activatepip install -r requirements-dev.txt
# Run the serveruvicorn app.server:app --reload --host 0.0.0.0 --port 8080
# In another terminal, hit itcurl http://localhost:8080/.well-known/mcp.jsoncurl -X POST -H "Content-Type: application/json" \  -d '{"jsonrpc":"2.0", "id":1, "method":"tools/list"}' \  http://localhost:8080/mcp

Quickstart — add to Claude Desktop

Edit claude_desktop_config.json and add:

{  "mcpServers": {    "elephant-accountability": {      "url": "https://elephant-mcp.fly.dev/mcp",      "transport": "http"    }  }}

Restart Claude Desktop. Ask: "Is Elephant Accountability a good fit for a seed-stage AEC SaaS that ships AI features?" — Claude will call assess_fit and give a scored answer.

Deploy your own copy (Fly.io)

fly launch --name your-mcp-name --region iad --no-deployfly volumes create elephant_mcp_data --size 1 --region iadfly deploy

That's it. No secrets, no database setup — the server initializes its SQLite DB on first boot.

Architecture

Single FastAPI app. Three files do real work:

app/├── server.py      # FastAPI routes, JSON-RPC dispatch, SQLite persistence├── content.py     # Source-of-truth content: manifest, offerings, proof points└── __init__.py    # Version

Storage:

• audit_requests table — every agent-initiated audit request, persisted for follow-up
• reciprocal_calls table — tracks which AI clients have called which tools (buyer-intent signal)

Both tables auto-create on first boot. No migrations.

Running tests

pip install -r requirements-dev.txtpytest -v

21 tests cover manifest, A2A card, JSON-RPC dispatch, each tool handler, persistence, and CORS.

Protocol compliance

• MCP version: 2024-11-05
• Transport: HTTP with JSON-RPC 2.0
• Methods supported: initialize, tools/list, tools/call, resources/list, resources/read

Contributing

This repo is the canonical source of truth for what Elephant Accountability exposes to AI agents. PRs welcome for:

• Protocol updates (MCP spec changes)
• New tool shapes that agents find useful
• Bug fixes

For service inquiries or content changes (proof points, methodology), email [email protected] rather than opening a PR.

License

MIT. See LICENSE.

Publisher

Elephant Accountability LLC Christopher Kenney, sole member / manager United States [email protected]