About cubic
cubic's hosted MCP server connects AI agents to cubic's GitHub pull request reviews and repository materials. It supports finding and triaging review issues, launching cloud reviews, and reading wikis, scans, and team learnings.
Key features
- 01Read pull request review findings
- 02Start cloud pull request reviews
- 03Update finding status and reply
- 04Browse repository wikis and scans
- 05Read team review learnings
- 06Inspect organization settings and members
Use cases
- 01Triage findings on a GitHub pull request
- 02Start a cloud review for a pull request
- 03Check scan issues and repository wikis
- 04Review team learnings for a repository
MCP support
The hosted MCP server exposes cubic review findings, repository documentation, codebase scans, team learnings, and organization information to MCP clients. Access and available tools follow the cubic account's permissions and subscription.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://www.cubic.dev/api/mcpclaude mcp add --transport http dev-cubic-cubic https://www.cubic.dev/api/mcp{"mcpServers":{"dev-cubic-cubic":{"type":"http","url":"https://www.cubic.dev/api/mcp"}}}Open in Cursor{"servers":{"dev-cubic-cubic":{"type":"http","url":"https://www.cubic.dev/api/mcp"}}}Open in VS Code{"mcpServers":{"dev-cubic-cubic":{"type":"http","url":"https://www.cubic.dev/api/mcp"}}}{"mcpServers":{"dev-cubic-cubic":{"serverUrl":"https://www.cubic.dev/api/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEcubic MCP server
cubic is an AI code review tool for GitHub pull requests. Its hosted MCP server connects your AI agent to cubic review findings, repository documentation, codebase scans, and team learnings.
What you can do
- Read open review findings on a pull request, including file locations, severity, and review status.
- Start a cloud review of a pull request.
- Resolve, reopen, or dismiss findings and record the outcome on the GitHub review thread.
- Read repository wikis and individual wiki pages.
- Browse codebase scan results, inspect issues, and update their triage status.
- Read the team's review learnings for a repository.
- Inspect organization subscriptions and members. Organization admins can also manage seats, roles, and supported billing settings.
MCP uses your cubic account's permissions. You need access to the relevant repository or organization; connecting the server does not grant additional access. Available tools depend on your permissions and subscription.
Connect
| Setting | Value |
|---|---|
| Server name | cubic |
| URL | https://www.cubic.dev/api/mcp |
| Transport | Streamable HTTP |
| Authentication | OAuth (recommended) or a personal API key |
Use the URL exactly as shown, including www. The server is hosted by cubic, so no local server installation is needed.
GitHub Copilot in VS Code
- Open the Command Palette and run MCP: Add Server.
- Choose HTTP and enter
https://www.cubic.dev/api/mcp. - Name the server
cubicand choose a user or workspace configuration. - Run MCP: List Servers, start cubic, and complete the browser OAuth sign-in.
- Use Copilot's agent mode with the cubic tools enabled.
You can also add this portable configuration to .mcp.json in your project. If the file already exists, add the cubic entry inside its mcpServers object:
{
"mcpServers": {
"cubic": {
"type": "http",
"url": "https://www.cubic.dev/api/mcp"
}
}
}
Cursor
Add the following to ~/.cursor/mcp.json for all projects, or .cursor/mcp.json for one project, then connect cubic in MCP settings and complete OAuth:
{
"mcpServers": {
"cubic": {
"url": "https://www.cubic.dev/api/mcp"
}
}
}
Claude Code
claude mcp add --transport http --scope user cubic https://www.cubic.dev/api/mcp
Open Claude Code, run /mcp, choose cubic, and complete the browser OAuth flow.
Codex CLI
codex mcp add cubic --url https://www.cubic.dev/api/mcp
codex mcp login cubic
codex mcp list
Scripts, CI, and clients without OAuth
Generate a personal API key in Settings > Integrations > MCP. Send it as a bearer token in the Authorization header and skip the client's OAuth login step.
For VS Code extension-host sessions, use an input in .vscode/mcp.json so the key is not written into the workspace configuration. For Agent Host sessions, use OAuth or a portable configuration with environment-based secrets; VS Code does not forward servers that require interactive input.
{
"inputs": [
{
"id": "cubic-api-key",
"type": "promptString",
"description": "cubic personal API key",
"password": true
}
],
"servers": {
"cubic": {
"type": "http",
"url": "https://www.cubic.dev/api/mcp",
"headers": {
"Authorization": "Bearer ${input:cubic-api-key}"
}
}
}
}
For scripts and CI, load the key from your secret store. Keys are personal and carry the creating user's permissions. You can revoke or regenerate a key from the same settings page.
Use it
After connecting, try these prompts, replacing the repository and PR with ones you can access:
- "Use cubic to list the open review issues on https://github.com/OWNER/REPO/pull/NUMBER."
- "Start a cubic review of PR #42 in acme/backend."
- "List wiki pages for acme/backend and show me the authentication system page."
- "Show codebase scan issues in acme/backend with severity at least 7."
- "What review learnings apply to acme/backend?"
get_pr_issues reads findings and review status; it does not start a review. An empty findings list means the current commit has no open findings only when its review completed and no newer commits exist. Use trigger_pr_review to start a cloud review.
Use PR issue IDs from get_pr_issues with update_pr_issue_status; this updates the GitHub review thread and queues an attributed reply. Use codebase scan issue IDs from get_scan with get_issue or update_issue_status.
For local code reviews, use the cubic CLI. cubic skills add workflows for investigating and fixing findings from your agent.
Documentation and support
- Full MCP documentation, available tools, and troubleshooting
- VS Code MCP setup and configuration
- cubic website
- Official MCP Registry entry
- Contact cubic support
This repository contains the public documentation and registry metadata for cubic's hosted MCP service. The registry name is dev.cubic/cubic and its metadata is in server.json.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.