Sandbox
@makeplane/plane-mcp-server

MCP server for Plane workspace operations

This server plugs Plane into any MCP-capable agent so the agent can read and change Plane projects, work items, cycles, and other workspace objects. It supports local stdio, remote HTTP, and SSE transports, with either OAuth or API key authentication.

312 stars172 forksPythonUpdated 7d ago
Who it's for

Builders who want their agent to manage Plane work items and projects.

What it delivers

You can ask your agent to inspect and update Plane without leaving your MCP client.

What it does

30 resource tools

Exposes one tool per Plane resource, covering 204 operations through an action parameter.

Multiple transports

Runs over local stdio, hosted HTTP, and deprecated SSE for backward compatibility.

OAuth and API key auth

Supports OAuth for hosted use and API key or personal access token flows for direct access.

PQL search support

Lets list, count, and search actions use Plane Query Language filters and grouping.

Self-documenting tool catalog

Each tool description lists its supported actions and required or optional parameters at call time.

Offline test suite

Includes tests that validate tool behavior against stand-in clients without network access.

How to get it

  1. 1Runs as a subprocess of your MCP client. Configuration as shown above; needs…
    PLANE_API_KEY=... PLANE_WORKSPACE_SLUG=... uvx plane-mcp-server stdio
  2. 2To onboard a new client without a release, append patterns
    export PLANE_OAUTH_ALLOWED_REDIRECT_URIS="https://newclient.com/cb,https://other.app/oauth/*"
  3. 3Structured JSON. Each tool call logs its name, duration, status and — when available —…
    export LOG_USER_INFO=false    # also log the display name (PII);
    export LOG_PAYLOADS=false    # keep request payloads out of logs; default true
  4. 4Only the OAuth and PAT transports carry a display name; stdio is unaffected.
    git clone https://github.com/makeplane/plane-mcp-server
    cd plane-mcp-server
    uv pip install -e ".[dev]"
  5. 5Run the server against a workspace
    PLANE_API_KEY=... PLANE_WORKSPACE_SLUG=... python -m plane_mcp stdio
    python -m plane_mcp http            # port 8211
  6. 6Tests, format, lint
    pytest                              # no network or credentials needed
    ruff format plane_mcp/ tests/       # line length 120
    ruff check plane_mcp/ tests/        # rules E, F, I, UP, B

README

Plane MCP Server

A Model Context Protocol server for Plane. Gives an AI agent tools to read and manage projects, work items, cycles, modules, releases, customers and more.

Built on FastMCP and the official plane-sdk.

  • 30 tools, one per Plane resource, covering 204 operations
  • Local or remote — stdio, streamable HTTP, SSE
  • OAuth or API key authentication

Quick start

Get an API key from Plane: Workspace Settings → API tokens.

Add this to your MCP client's configuration:

{
  "mcpServers": {
    "plane": {
      "command": "uvx",
      "args": ["plane-mcp-server", "stdio"],
      "env": {
        "PLANE_API_KEY": "<your-api-key>",
        "PLANE_WORKSPACE_SLUG": "<your-workspace-slug>"
      }
    }
  }
}

uvx needs no install step. Requires Python 3.10+.

For a self-hosted Plane, add "PLANE_BASE_URL": "https://plane.example.com".

Transports

stdio — local

Runs as a subprocess of your MCP client. Configuration as shown above; needs PLANE_API_KEY and PLANE_WORKSPACE_SLUG.

PLANE_API_KEY=... PLANE_WORKSPACE_SLUG=... uvx plane-mcp-server stdio

HTTP with OAuth — hosted

https://mcp.plane.so/http/mcp

The OAuth flow is handled on connect; no credentials in your config. For clients without native remote MCP support, bridge with mcp-remote:

{
  "mcpServers": {
    "plane": {
      "command": "npx",
      "args": ["mcp-remote@latest", "https://mcp.plane.so/http/mcp"]
    }
  }
}

Requires Node.js 22+.

HTTP with a personal access token — hosted

https://mcp.plane.so/http/api-key/mcp

HeaderValue
AuthorizationBearer <PAT>
X-Workspace-slug<workspace-slug>
{
  "mcpServers": {
    "plane": {
      "command": "npx",
      "args": ["mcp-remote@latest", "https://mcp.plane.so/http/api-key/mcp"],
      "headers": {
        "Authorization": "Bearer <PAT>",
        "X-Workspace-slug": "<workspace-slug>"
      }
    }
  }
}

SSE — deprecated

https://mcp.plane.so/sse is maintained for backward compatibility only. Use an HTTP transport instead.

Tools

The server advertises 30 tools, one per resource. Each takes an action parameter that selects the operation:

workitem(action="create", project_id=..., name="Fix login")
workitem(action="list", project_id=..., pql='state__group = "started"')
cycle(action="archive", project_id=..., cycle_id=...)

Every tool's description lists its actions with their required and optional parameters, so the catalogue is self-documenting at call time.

Full tool and action reference

Querying work items

List, count and search accept PQL, Plane's query language:

workitem(action="list", project_id=..., pql='state__group = "started" AND priority = "urgent"')
workitem(action="count", pql='assignees__id = "<member id>"', group_by="state_id")

Call get_pql_reference for the full syntax, operators and worked examples.

Upgrading from the per-operation tools

Earlier releases exposed one tool per API operation. Existing integrations keep working: 169 of those 177 names still resolve to the consolidated tool, so a saved prompt or script calling create_work_item or list_cycles needs no change. They are no longer advertised, and they keep the parameter names they shipped with (work_item_id, not workitem_id).

Seven names chose between two operations with a parameter (manage_project_archive(archive=False)), which one tool-and-action pair cannot reproduce; calling one tells you its replacement. get_pql_reference is unchanged.

Configuration

Authentication

VariableRequired forPurpose
PLANE_API_KEYstdioAPI key
PLANE_WORKSPACE_SLUGstdioTarget workspace
PLANE_BASE_URLoptionalPlane API URL (default https://api.plane.so)

The remote transports carry credentials in the connection — the OAuth flow or the PAT headers — and need none of these.

Self-hosting the server itself:

VariablePurpose
PLANE_INTERNAL_BASE_URLInternal URL for server-to-server calls, preferred over PLANE_BASE_URL
REDIS_URLOAuth token storage as one connection URL (redis:// or rediss:// for TLS); wins over host/port
REDIS_HOST / REDIS_PORTOAuth token storage; falls back to in-memory
PLANE_OAUTH_PROVIDER_*OAuth client credentials and base URL
MCP_PATH_PREFIXPath prefix for the HTTP routes, when mounted behind a proxy — /plane serves /plane/http/mcp

OAuth redirect URIs

The OAuth transports validate each client's redirect URI against an allowlist. Common clients (Cursor, VS Code, Claude.ai, ChatGPT connectors, localhost) are allowed by default.

To onboard a new client without a release, append patterns:

export PLANE_OAUTH_ALLOWED_REDIRECT_URIS="https://newclient.com/cb,https://other.app/oauth/*"

* matches any port, path segment or subdomain. Keep the host pinned and wildcard only the port or path.

Logging

Structured JSON. Each tool call logs its name, duration, status and — when available — an opaque user id and the workspace slug.

export LOG_USER_INFO=false    # also log the display name (PII);
export LOG_PAYLOADS=false    # keep request payloads out of logs; default true

Only the OAuth and PAT transports carry a display name; stdio is unaffected.

Development

git clone https://github.com/makeplane/plane-mcp-server
cd plane-mcp-server
uv pip install -e ".[dev]"

Run the server against a workspace:

PLANE_API_KEY=... PLANE_WORKSPACE_SLUG=... python -m plane_mcp stdio
python -m plane_mcp http            # port 8211

Tests, format, lint:

pytest                              # no network or credentials needed
ruff format plane_mcp/ tests/       # line length 120
ruff check plane_mcp/ tests/        # rules E, F, I, UP, B

The suite runs fully offline — every action of every resource is executed against a stand-in that binds each call against the genuine plane-sdk signature. See plane_mcp/tools/README.md.

Live integration tests are skipped unless you point them at a running server:

export PLANE_TEST_API_KEY=... PLANE_TEST_WORKSPACE_SLUG=...
export PLANE_TEST_MCP_URL=http://localhost:8211    # optional; this is the default
pytest tests/test_integration.py -v

They write real data to that workspace.

Repository layout

PathContents
plane_mcp/__main__.pyentry point; picks the transport from argv[1]
plane_mcp/server.pyone factory per transport
plane_mcp/client.pyresolves credentials into a plane-sdk client
plane_mcp/auth/OAuth provider and header auth
plane_mcp/tools/the tool surface: one module per Plane resource
plane_mcp/toolkit/shared building blocks for the tool surface
plane_mcp/pql_reference.pyPQL syntax reference served to models

Contributing

Pull requests welcome. Please run pytest and ruff check before submitting; new tools should come with the invariants described in plane_mcp/tools/README.md.

See CONTRIBUTING.md and CODE_OF_CONDUCT.md.

Migrating from the Node.js server

@makeplane/plane-mcp-server (Node.js) is deprecated and unmaintained. This Python implementation replaces it.

Node.jsPython
PLANE_API_KEYPLANE_API_KEY
PLANE_API_HOST_URLPLANE_BASE_URL
PLANE_WORKSPACE_SLUGPLANE_WORKSPACE_SLUG

Replace the command and args with the stdio configuration in Quick start.

License

MIT — see LICENSE.

Files in the repo

Repository payload18 top-level entries
  • .github
  • plane_mcp
  • tests
  • .dockerignore
  • .editorconfig
  • .env.test
  • .gitignore
  • CLAUDE.md
  • CODE_OF_CONDUCT.md
  • CONTRIBUTING.md
  • Dockerfile
  • glama.json
  • LICENSE
  • mise.toml
  • pyproject.toml
  • README.md
  • SECURITY.md
  • uv.lock

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More connectors

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

86k

High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static binary, zero dependencies.

43k

Universal provider proxy for OpenAI Codex & Claude Code — use any LLM (Claude, Gemini, Grok, DeepSeek, Ollama…) with Codex CLI, App, SDK, and Claude Code

14k
okf-memory/
okf-agent-memory

Git-native persistent memory for AI coding agents. Implements Google OKF v0.2 with sub-300µs in-memory BM25 search, embedded MCP server, and progressive disclosure. Slashes token bloat by 80% with zero external databases or dependencies. Built in pure Go.

547
tirth8205/
code-review-graph

Local-first code intelligence graph for MCP and CLI. Builds a persistent map of your codebase so AI coding tools read only what matters, with benchmarked context reductions on reviews and large-repo workflows.

31k
2akouwu/
reverify

Stop your AI from making things up — it proposes, deterministic tools decide, every claim checked against ground truth with evidence. Grounded facts and context survive resets. Reverse engineering is the proving ground. MCP server + CLI.

1.1k