Outset MCP

Connect AI agents to Outset — the AI-moderated research platform — via the Model Context Protocol.

The Outset MCP server is remote: there is nothing to install or run. Point your MCP client at the endpoint below and log in with your Outset account.

https://api.outset.ai/mcp/

• Transport: Streamable HTTP (stateless JSON-RPC)
• Auth: OAuth 2.1 with PKCE + Dynamic Client Registration — compliant clients connect with zero pre-registration
• Full developer guide: Outset MCP — Integration Guide for Developers (OAuth flow, scopes, DCR details, testing)

Connect from your client

Claude (claude.ai / Desktop)

Settings → Connectors → Add custom connector, name it Outset, paste https://api.outset.ai/mcp/, then Connect to go through the OAuth consent flow. (Custom connectors require a Pro, Max, Team, or Enterprise plan.)

Claude Code

claude mcp add --transport http outset https://api.outset.ai/mcp/

Or install as a plugin from this repo:

claude plugin marketplace add Outset-AI/outset-mcp
claude plugin install outset@outset-mcp

Cursor

Or Settings → MCP → Add Server and paste the URL — Cursor handles registration and OAuth automatically.

ChatGPT

Enable Developer Mode under Settings → Apps & Connectors → Advanced settings, then Create a connector with https://api.outset.ai/mcp/ as the MCP server URL and OAuth as the auth method.

VS Code

code --add-mcp '{"name":"outset","type":"http","url":"https://api.outset.ai/mcp/"}'

Gemini CLI

gemini extensions install https://github.com/Outset-AI/outset-mcp

Codex

Add to ~/.codex/config.toml (or .codex/config.toml in your project):

[mcp_servers.outset]
url = "https://api.outset.ai/mcp/"

Then authenticate with:

codex mcp login outset

Anything else

Any MCP client that supports remote servers (Streamable HTTP + OAuth) works. For stdio-only clients, bridge with mcp-remote:

npx -y mcp-remote https://api.outset.ai/mcp/

To explore the tool list interactively:

npx @modelcontextprotocol/inspector https://api.outset.ai/mcp/

What your agent can do

• Create and edit studies — projects, sections, questions, screeners, display logic, quotas, concepts
• Configure and launch recruitment — sample size, reward, demographic filters (launching may incur fees and is gated behind the separate studies:launch scope)
• Query results — reports, answers, topline insights, category summaries, highlight clips
• Look up org context — workspaces, projects, organization info

The live tools/list response is the source of truth — new tools ship regularly. Scopes and org roles are enforced server-side; see the integration guide for the full scope table.

What's in this repo

Manifests for agent directories and marketplaces — the server itself is not here (it's remote):

File	Purpose
.claude-plugin/plugin.json + .mcp.json	Claude Code plugin wrapping the remote server
.claude-plugin/marketplace.json	Makes this repo installable as a Claude Code plugin marketplace
.cursor-plugin/marketplace.json	Same, for Cursor's plugin system
gemini-extension.json	Gemini CLI extension (gemini extensions install <repo url>)
server.json	Manifest for the official MCP registry