Skip to content

Connect Nibvine to coding agents and AI clients.

Project context, notes, tasks, recent changes and preview-then-commit writes are all served from one remote MCP endpoint. Use the connect flow for a client-specific install pack, or wire it up by hand.

Supported clients

These are the install surfaces Nibvine packages today for discovery and fast setup.

Claude Code

CLI-ready remote MCP setup for agentic coding and review loops.

Install surface: Nibvine preset + project-scoped token

Official docs

Cursor

Repo-native MCP config plus one-click install links for fast onboarding.

Install surface: Cursor MCP config or deeplink

Official docs

Codex

Remote MCP setup with an exact `codex mcp add` command and TOML snippet.

Install surface: Codex CLI command or config.toml

Official docs

VS Code

Workspace-safe MCP config for Copilot agent mode with token prompts.

Install surface: .vscode/mcp.json

Official docs

Any MCP client

Fallback payload, smoke test, manifest link, and OpenAPI discovery links.

Install surface: Generic MCP JSON + curl smoke test

Official docs

Recommended setup path

  1. Step 1

    Create or reuse a project-scoped Nibvine token.

  2. Step 2

    Pick a client preset or use the generic MCP payload.

  3. Step 3

    Call get_context_pack first so the agent starts with project context instead of raw notes.

  4. Step 4

    For writes, use preview_create_note, preview_create_task, or preview_update_task before commit_write_action.

Minimal remote MCP check

A project-scoped bearer token is enough to confirm the server is reachable and listing tools.

smoke test
curl -s https://nibvine.com/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <NIBVINE_TOKEN>" \
  -H "MCP-Protocol-Version: 2025-06-18" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'