Use the CVPitch MCP server from your favourite AI app

The CVPitch MCP server is a remote Model Context Protocol server, so you can work with your candidate submissions from the AI app you already use: Claude, ChatGPT, Microsoft Copilot, Gemini, Le Chat, Perplexity, Cursor, VS Code and any other app that supports remote MCP. You add one address, sign in to CVPitch and choose what the app may do. This page has the steps for each app.

The address to add

Every app uses the same server address:

https://cvpitch.app/mcp

Most apps sign in with OAuth. You don't need a client ID or secret: the app registers itself, then opens a CVPitch page where you sign in, check the permissions and choose Allow access. A few apps can't sign in this way and use an API key instead (see apps that need an API key).

You need a CVPitch workspace. Start a free trial if you don't have one yet.

What your assistant can do

Your assistant can use everything you do in CVPitch, except approving. Ask things like "Which submissions need review?", "Rewrite the summary for submission 1a2b3c4d to suit a fintech client", or "Put the three approved Java developers on a shortlist and share it with the client". The tools cover:

• Submissions: search and list them, edit facts, compare versions, restore an earlier revision, reuse a CV for another client, and delete one when you ask.
• AI writing: summaries, rewrites, proofreading, translation, achievement prompts and descriptors. Suggestions wait for you to accept them.
• Job match and pitches: match a CV against a job description and save the pitch.
• Previews and exports: create a draft preview, export a submission a recruiter has approved, and get a short-lived download link.
• Shortlists and client sharing: build and order shortlists, create, extend and revoke client share links, and see client views and feedback.
• Interview notes: see and remove interview transcripts imported in CVPitch.
• Workspace: usage and credits, templates, client formats, email intake, your talent page and saved AI prompts. Owner-only settings still need an owner.
• Free helpers: anonymise text, find personal data, check keywords against a job description, check wording for bias, and the time-saving and cost calculators.
• Uploads: a short-lived link opens the CVPitch upload screen, so CV files never pass through the chat.

The assistant can never approve a submission: a person approves every one in CVPitch, and exports only use an existing approval. Steps where a person vouches for something also stay in the app: resolving review flags, accepting a client template, confirming a candidate's identity or consent, and importing interview transcripts.

When you connect, you choose the permissions. Sharing with clients, ATS pushes and the public talent page need shares:write, workspace settings need workspace:write, and permanent deletion needs submissions:delete. These are off unless you tick them. Tools that work with candidate content, such as AI writing and job match, also need assistant sharing and candidate:read. If you connected before 13 October 2026, disconnect and connect again to get the full set of tools.

Candidate names and CV content are not sent to the assistant unless a workspace owner turns on assistant sharing and you grant the candidate:read permission. Even then, the raw CV text, recruiter notes and contact details are withheld, and names are masked if your workspace's sharing settings replace them. Masking is automatic and approximate, so it isn't a guarantee of anonymity. Whatever a tool returns is processed by your AI app's provider, so read how we handle AI before turning sharing on.

Claude

Works in Claude on the web, the desktop app and mobile.

1. Pro and Max: open Customize → Connectors (or Settings → Connectors) and choose + Add → Add custom connector. Team and Enterprise: an owner adds the connector once under Organization settings → Connectors, then each person connects.
2. Name it CVPitch, paste https://cvpitch.app/mcp and leave the OAuth client fields empty.
3. Choose Connect, sign in to CVPitch and choose Allow access.
4. In a chat, turn on CVPitch from the tools menu.

Free Claude accounts can add one custom connector.

ChatGPT

1. Open Settings → Apps (older menus say Apps & Connectors) → Advanced settings and turn on Developer mode. In Business and Enterprise workspaces an admin may need to allow this, or create the app under Workspace settings → Apps.
2. Choose Create. Name: CVPitch. MCP server URL: https://cvpitch.app/mcp. Authentication: OAuth.
3. Confirm you trust the app, choose Create, then sign in to CVPitch and choose Allow access.
4. In a new chat, pick CVPitch from the + menu.

Custom MCP apps are offered on paid ChatGPT plans, and which plans can use write actions changes over time. ChatGPT can't use API keys, only the sign-in above.

Microsoft Copilot

Add CVPitch as a tool to a Copilot Studio agent, then publish the agent to Microsoft Teams or Microsoft 365 Copilot as usual.

1. In Copilot Studio, open your agent and go to Tools → Add a tool → New tool → Model Context Protocol.
2. Server name: CVPitch. Server URL: https://cvpitch.app/mcp. Authentication: OAuth 2.0 with Dynamic discovery. Choose Create.
3. Choose Create a new connection, sign in to CVPitch, choose Allow access, then Add to agent.

Your organisation's Power Platform data policies may need to allow the new connector. For GitHub Copilot in VS Code, see developer tools.

Gemini

Gemini Enterprise: a Google Cloud admin adds a data store of type Custom MCP Server with the URL https://cvpitch.app/mcp and OAuth. Each person then signs in to CVPitch the first time they use it. Custom MCP servers are a preview feature in Gemini Enterprise and may be blocked by an organisation policy until an admin allows them.

Gemini CLI:

gemini mcp add --transport http cvpitch https://cvpitch.app/mcp

Then run /mcp auth cvpitch inside Gemini CLI to sign in.

Mistral Le Chat

1. Open Connectors → + Add Connector and choose the Custom MCP Connector tab.
2. Name it CVPitch, enter https://cvpitch.app/mcp as the server URL and choose Connect.
3. Sign in to CVPitch and choose Allow access.

Only Le Chat admins can add custom connectors; on personal plans you are the admin.

Perplexity

1. Open Settings → Connectors → + Custom connector and choose Remote.
2. Name it CVPitch, enter https://cvpitch.app/mcp and choose OAuth.
3. Sign in to CVPitch and choose Allow access.

Custom connectors need a Pro, Max or Enterprise plan. On Enterprise, an admin must first allow members to add custom connectors.

Cursor, VS Code and developer tools

Claude Code:

claude mcp add --transport http cvpitch https://cvpitch.app/mcp

Then run /mcp inside Claude Code and sign in.

Cursor: add this to ~/.cursor/mcp.json, then open Cursor Settings → MCP and choose Connect next to CVPitch.

{ "mcpServers": { "cvpitch": { "url": "https://cvpitch.app/mcp" } } }

VS Code with GitHub Copilot: add this to .vscode/mcp.json (or run MCP: Open User Configuration), choose Start above the entry and sign in. On Copilot Business and Enterprise, an organisation admin must allow MCP servers.

{ "servers": { "cvpitch": { "type": "http", "url": "https://cvpitch.app/mcp" } } }

OpenAI Codex CLI:

codex mcp add cvpitch --url https://cvpitch.app/mcp
codex mcp login cvpitch

Zed: add "context_servers": { "cvpitch": { "url": "https://cvpitch.app/mcp" } } to your settings; Zed opens the sign-in the first time the agent uses it.

Cline: open the MCP Servers icon → Remote Servers, enter CVPitch and the URL, choose Streamable HTTP and add it.

Goose: run goose configure → Add Extension → Remote Extension (Streamable HTTP) and enter the URL.

Windsurf, Roo Code and LM Studio use an API key in a header, for example in Windsurf's mcp_config.json:

{ "mcpServers": { "cvpitch": { "serverUrl": "https://cvpitch.app/mcp", "headers": { "Authorization": "Bearer cvp_live_..." } } } }

Roo Code uses "type": "streamable-http" with "url", and LM Studio uses "url", each with the same headers.

JetBrains AI Assistant and other apps that only run local servers can use the open-source mcp-remote bridge, which needs Node.js. In JetBrains, add it under Settings → Tools → AI Assistant → Model Context Protocol (MCP):

{ "mcpServers": { "cvpitch": { "command": "npx", "args": ["-y", "mcp-remote", "https://cvpitch.app/mcp"] } } }

n8n: add an MCP Client tool to an AI Agent node with the URL, transport HTTP Streamable and Header Auth using an API key.

Apps that need an API key

Apps without OAuth sign-in send a fixed Authorization: Bearer header. A workspace owner creates the key in CVPitch under Integrations → REST API, chooses only the permissions it needs and ticks Also allow MCP clients. The key is shown once. Treat it like a password and revoke it from the same screen when you no longer need it.

Managing connections

Every connection appears under Integrations in CVPitch, with the app, the person who connected it, its permissions and when it was last active. Owners see every connection in the workspace and can disconnect any of them; the app loses access immediately. A connection that isn't used for 30 days signs out, and every connection asks you to sign in again after 90 days.

If an app can't connect, disconnect it, try again, and contact support with the app's name. Menu names in these apps change often; the server address stays the same. Technical details, scopes and limits are on the developers page.