AI Assistants (MCP)

The Peerdom MCP server lets any AI assistant answer questions about your organization — your people, roles, circles, who holds what, where the gaps are — in plain language. It speaks the Model Context Protocol (MCP), an open standard for connecting AI assistants to tools and data, so it works with the assistant you already use rather than tying you to one vendor.

You connect it with your existing Peerdom API key. Nothing to install on your side, and you can be up and running in about a minute.

The MCP server supports both reading and writing. Assistants can ask about your organization and, with the same API key, make changes (create, update, and remove peers, roles, and groups). Your API key controls what assistants can do; revoke it any time to disconnect every assistant at once.

What you can ask

Once connected, you can ask your assistant questions like:

“What organization am I connected to?”
“Which roles in the Marketing circle are currently vacant?”
“Who holds the Finance Lead role, and what’s their contribution?”
“What roles does Maria hold, and in which circles are they?”
“Who is carrying the most roles right now — is anyone over-committed?”
“Show me our organization structure as an outline.”

Behind the scenes the assistant uses a set of tools: read tools to list and inspect peers, roles, and circles, find vacant roles, and summarize workload, plus write tools to create, update, and remove peers, roles, and groups. The assistant chooses which to call based on your question and answers in natural language.

What every client needs

Every MCP client connects to the same endpoint with the same header:

Endpoint:  https://mcp.peerdom.org/mcp
Header:    X-Api-Key: <your-api-key>

Get your API key

Go to Settings > My Data > API Keys in Organization Settings and click Create a key. Only users with Owner access rights can create and manage API keys. This is the same key used by the REST API — one key per organization.

Your API key grants full read and write access to your organization's data. Treat it like a password: never paste it into a public repository or share it in a screenshot, and store it only somewhere your assistant client can reach it. You can revoke a key at any time from Organization Settings, which immediately disconnects every assistant using it.

Set up your client

Pick the assistant you use. Each one needs the endpoint and your API key in its configuration.

Claude Desktop

Claude Desktop reaches remote servers through a small bridge called mcp-remote. Open Settings > Developer > Edit Config and add:

{
  "mcpServers": {
    "peerdom": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.peerdom.org/mcp", "--header", "X-Api-Key:${PEERDOM_API_KEY}"],
      "env": { "PEERDOM_API_KEY": "your-api-key" }
    }
  }
}

Save the file and restart Claude Desktop. The Peerdom tools appear under the connectors icon.

Note the header has no space after the colon (X-Api-Key:${PEERDOM_API_KEY}). The bridge splits the argument on the first space, so keeping the value attached avoids a common connection error.

Claude Code

From your terminal:

claude mcp add --transport http peerdom https://mcp.peerdom.org/mcp \
  --header "X-Api-Key: your-api-key"

Run claude mcp list to confirm it’s connected.

Cursor

Open Settings > MCP > Add new MCP server, or edit ~/.cursor/mcp.json directly:

{
  "mcpServers": {
    "peerdom": {
      "url": "https://mcp.peerdom.org/mcp",
      "headers": { "X-Api-Key": "your-api-key" }
    }
  }
}

VS Code / GitHub Copilot

In agent mode, add a .vscode/mcp.json file to your workspace:

{
  "servers": {
    "peerdom": {
      "type": "http",
      "url": "https://mcp.peerdom.org/mcp",
      "headers": { "X-Api-Key": "${input:peerdom_key}" }
    }
  }
}

VS Code prompts for the key the first time and stores it securely.

On GitHub Copilot Business and Enterprise, MCP is gated by an administrator and is off by default. Ask your Copilot admin to enable MCP servers if you don't see the option.

ChatGPT

ChatGPT supports custom MCP servers in Developer Mode (beta). Enable it under Settings > Connectors > Advanced > Developer mode, then add a connector pointing at https://mcp.peerdom.org/mcp.

For programmatic use, the OpenAI Responses API accepts our server directly as an mcp tool:

{
  "type": "mcp",
  "server_label": "peerdom",
  "server_url": "https://mcp.peerdom.org/mcp",
  "headers": { "X-Api-Key": "your-api-key" }
}

The ChatGPT consumer app's published connectors require OAuth, which the Peerdom server does not offer yet — so use Developer Mode or the Responses API for now. OAuth support is coming (see What works today).

Gemini CLI

Edit ~/.gemini/settings.json:

{
  "mcpServers": {
    "peerdom": {
      "httpUrl": "https://mcp.peerdom.org/mcp",
      "headers": { "X-Api-Key": "your-api-key" }
    }
  }
}

Local and Swiss-hosted models

Because MCP is an open standard, you can keep everything on infrastructure you trust. Point a local MCP client such as LM Studio at the server:

{
  "mcpServers": {
    "peerdom": {
      "url": "https://mcp.peerdom.org/mcp",
      "headers": { "X-Api-Key": "your-api-key" }
    }
  }
}

Pair an MCP-capable client with a local open-weight model (for example Qwen3, GLM-4.5, or Gemma) for a fully offline stack, or with a Swiss-sovereign model API such as Infomaniak for an EU/CH-hosted setup. The assistant runs where you choose; only your API key reaches Peerdom.

Want to choose the model yourself? The Bring your own LLM guide walks through a developer setup with Infomaniak's Swiss-hosted, OpenAI-compatible API as a worked example.

What works today

The server authenticates with an API-key header, which every developer- and IDE-class assistant supports. Cloud chat apps that only accept sign-in with OAuth aren’t reachable yet — that’s a single planned feature, not a limitation of your data.

Client	Works today	How	Notes
Claude Desktop	✅	`mcp-remote` bridge	Native connector UI has no header field — use the config above
Claude Code	✅	Direct	Cleanest fit
Cursor	✅	Direct	`headers` in `mcp.json`
VS Code / Copilot	✅	Direct	Business/Enterprise: admin must enable MCP
ChatGPT — Developer Mode	✅	Direct	Beta; confirm the custom-header field in the UI
OpenAI Responses API	✅	Direct	Native `mcp` tool
Gemini CLI	✅	Direct	`httpUrl` + `headers`
LM Studio / local agents	✅	Direct	Keep the model local or Swiss-hosted
claude.ai (web)	⏳	—	Custom connectors require OAuth — coming soon
ChatGPT — published connectors	⏳	—	Require OAuth — coming soon
Gemini consumer app	❌	—	No custom MCP support

Verify it works

Ask your assistant: “What organization am I connected to?” It should reply with your organization’s name — that confirms the key, the connection, and the data path all work.

To test without a client, send a request directly:

curl -s -X POST https://mcp.peerdom.org/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -H 'X-Api-Key: your-api-key' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_organization","arguments":{}}}'

A missing or invalid key returns 401 Missing X-Api-Key header.

Frequently asked questions

Can the assistant change my organization?

Yes. With your API key, an assistant can create, update, and remove peers, roles, and groups, set custom field values, assign role holders, and more. Most AI clients (Claude Desktop, Claude Code, Cursor, etc.) surface every tool call before invoking it, so you can confirm or cancel each change. Some operations are reversible (deleting a peer is a soft delete that keeps the person in your archive), while removing a role or group is permanent. Use an assistant you trust, and revoke the API key anytime to disconnect every assistant at once.

Does Peerdom store my API key?

No. Your key is passed through with each request to the Peerdom API and is never stored on the MCP server. Revoke it anytime in Organization Settings to disconnect every assistant at once.

Which assistants can I use?

Any MCP client that lets you set a request header — see What works today. You’re not tied to a single vendor: bring the assistant you already trust.

Do I need a specific Peerdom plan?

API access is available on paid plans. Contact Peerdom support for details on your plan.

What data can the assistant see?

Whatever your API key can read: your organization’s peers, roles, circles, holders, vacancies, and workload. It’s scoped to the one organization the key belongs to.

Bring your own LLM, choose your own model (Infomaniak, local, sovereign)
Agent-Readable Export (OKF), publish your organization for any AI agent to read
API and Webhooks, build custom integrations with the REST API
Integrations Overview, all available integrations at a glance
Access Rights, understand who can create API keys
Organization Settings, where to find and revoke API keys
Your org chart can answer questions now, the story behind this integration