# echoloc MCP server — company technographics for AI agents

Remote [Model Context Protocol](https://modelcontextprotocol.io) server for the
[echoloc](https://echoloc.ai) company-intelligence dataset: search 760,000+
companies by the technologies they use — with **direction of change** (what
each company is *adopting*, *replacing*, and *evaluating*), hiring signals, and
firmographics. Signals are extracted from millions of public job posts, so the
data covers backend, data, cloud, and AI tooling that never appears in website
source.

- **Endpoint:** `https://api.echoloc.ai/mcp` (Streamable HTTP, stateless)
- **Registry name:** [`ai.echoloc/company-technographics`](https://registry.modelcontextprotocol.io/v0/servers?search=technographics)
- **Auth:** `X-API-Key` header (also accepts `Authorization: Bearer`)
- **Free key:** 100 requests/month, instant — [sign up](https://echoloc.ai/auth?mode=signup&returnTo=%2Fapp%2Fapi) and copy it from [echoloc.ai/app/api](https://echoloc.ai/app/api)

## Tools

| Tool | What it does |
| --- | --- |
| `search_companies_by_technologies` | Find companies using given technologies (match `any`/`all`), ranked by usage. Case-insensitive, 10,000+ technologies tracked. |
| `get_company_by_domain` | Resolve a domain (e.g. `walgreens.com`) to a full profile: firmographics, hiring stats, tech stack with adoption context. |
| `list_technologies` | The full technology catalog with per-technology company counts. |

All tools are read-only (`readOnlyHint: true`). Discovery (`initialize`,
`tools/list`) requires no key; tool calls consume 1 credit each.

## Quick start

### Claude Code

```bash
claude mcp add --transport http echoloc https://api.echoloc.ai/mcp \
  --header "X-API-Key: YOUR_API_KEY"
```

### Cursor / any MCP client

```json
{
  "mcpServers": {
    "echoloc": {
      "url": "https://api.echoloc.ai/mcp",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}
```

Then ask things like:

> *Which financial-services companies are adopting Snowflake right now?*
>
> *Pull the tech stack and hiring signals for walgreens.com.*
>
> *Which companies are actively replacing VMware?*

### stdio bridge (Claude Desktop, local runners)

For stdio-only MCP clients this repo ships a dependency-free bridge
([`mcp_stdio_bridge.py`](mcp_stdio_bridge.py)) that forwards JSON-RPC to the
hosted server:

```json
{
  "mcpServers": {
    "echoloc": {
      "command": "python3",
      "args": ["/path/to/mcp_stdio_bridge.py"],
      "env": { "ECHOLOC_API_KEY": "YOUR_API_KEY" }
    }
  }
}
```

Or via Docker:

```bash
docker build -t echoloc-mcp .
docker run -i --rm -e ECHOLOC_API_KEY=YOUR_API_KEY echoloc-mcp
```

## Raw JSON-RPC example

```bash
curl https://api.echoloc.ai/mcp \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_companies_by_technologies","arguments":{"technologies":["Snowflake"],"match":"any","page_size":10}}}'
```

## What makes the data different

Most technographics are scraped from website source and can only see frontend
tools. echoloc reads what companies say in their own hiring:

- **Direction of change** — `tech_adopting`, `tech_replacing`, `tech_evaluating`
  arrays on every profile, not just a static "uses X" flag.
- **Hiring context** — active jobs, hiring velocity, departments, leadership
  hires.
- **Depth** — 360,000+ companies with detected stacks across 10,000+
  technologies, refreshed daily.

## Links

- Agent fast lane: [echoloc.ai/for-agents](https://echoloc.ai/for-agents/)
- REST API docs: [echoloc.ai/developers](https://echoloc.ai/developers) · [OpenAPI spec](https://api.echoloc.ai/openapi.json)
- LLM-readable summary: [llms.txt](https://echoloc.ai/llms.txt) · [llms-full.txt](https://echoloc.ai/llms-full.txt)
- Methodology: [echoloc.ai/methodology](https://echoloc.ai/methodology/)
- Research built on this dataset: [echoloc.ai/research](https://echoloc.ai/research/)

## Support

Questions, production keys (higher limits, bulk feeds, licensing):
**hello@echoloc.ai** — we reply within one business day.