plane-mcp
MCP server for Plane, the open-source project management tool, exposing a focused set of task and comment management tools for a single Plane project. Includes a setup wizard to bind a local folder to a Plane project and configure Claude Code or Codex.
README
plane-mcp
An MCP (Model Context Protocol) server for Plane, the open-source project management tool — plus a setup wizard that binds a project folder to a Plane project in one command.
One CLI, four subcommands:
plane-mcp(orplane-mcp setup) — interactively binds the current folder to a Plane project (writes MCP config for Claude Code / Codex, updates the agent memo).plane-mcp mcp— the MCP server itself (stdio).plane-mcp list— lists projects from your local registry.plane-mcp update— self-upgrade via pipx.
Design
Most Plane MCP servers expose dozens of tools and make the model pass
workspace and project on every call. This one is deliberately narrow:
one MCP server instance = one Plane project, fixed via environment
variables at startup. The model gets a small, task-focused tool surface and
can never write to the wrong project.
Tools (MCP)
| Tool | Purpose |
|---|---|
list_tasks(status_group?, state_id?, assignee_id?, limit?, cursor?) |
List tasks; status_group=backlog shows only the backlog. |
get_task(id_or_identifier) |
By UUID or PROJ-42 identifier. |
create_task(name, …) |
Create a task. |
update_task(issue_id, …) |
Patch any fields. |
change_status(issue_id, status_group OR state_id) |
Sugar over update. |
add_comment(issue_id, body_html) |
Post an HTML comment (sanitized via nh3). |
list_comments(issue_id) |
Read comments, incl. presigned URLs for inline images. |
add_comment_with_image(issue_id, body_html, image_path) |
Comment with an embedded image. |
upload_attachment(issue_id, file_path) |
Attach a local file. |
list_attachments(issue_id) / get_attachment_url(…) / delete_attachment(…) |
Manage attachments. |
read_attachment_image(issue_id, attachment_id) |
Return an image attachment as native MCP image content. |
list_states() / list_members() / list_labels() |
Lookups, cached 5 min. |
status_group ∈ backlog, unstarted, started, completed, cancelled.
Installation
brew install pipx # or: sudo apt install pipx
pipx ensurepath
pipx install git+https://github.com/babazulu/plane-mcp
Configure your projects
Create ~/.config/plane-mcp/projects.toml with one entry per Plane project
you work with:
[projects.MyProject]
id = "00000000-0000-0000-0000-000000000000" # Plane project UUID
api_key = "plane_api_..." # Plane API key (Workspace Settings → API tokens)
workspace_slug = "my-workspace"
base_url = "https://plane.example.com" # your Plane host (or https://app.plane.so)
Verify with:
plane-mcp list
Bind a project folder
cd /path/to/my/project
plane-mcp # interactive: pick the agent and the Plane project
Or fully flag-driven:
plane-mcp setup --agent codex --project MyProject
plane-mcp setup --agent claude --project MyProject --dir /other/path
What it does:
- Creates/merges
./.mcp.json(Claude Code) or./.codex/config.toml(Codex) — adds or updates theplane-<name>MCP server entry and leaves existing entries untouched. - Upserts an agent memo block in
CLAUDE.md(Claude) orAGENTS.md(Codex), bounded by<!-- plane-mcp:start --> … end -->markers. - Prints a first prompt to copy into the agent.
On the next launch in that folder, Claude Code or Codex picks up the
plane-<name> MCP server and can work with tasks.
MCP server configuration (ENV)
The MCP host (Claude Code / Codex) passes these from its config. If you run
plane-mcp mcp by hand, set them yourself:
PLANE_API_URL=https://plane.example.com
PLANE_API_KEY=...
PLANE_WORKSPACE_SLUG=...
PLANE_PROJECT_ID=<uuid>
Development
git clone https://github.com/babazulu/plane-mcp
cd plane-mcp
python3 -m venv .venv && . .venv/bin/activate
pip install -e .
plane-mcp list
plane-mcp setup --dir /tmp/test-folder --agent codex --project MyProject
Releases
- Bump
versioninpyproject.tomlandsrc/plane_mcp/__init__.py. - Commit and push (CI runs lint + build).
- Tag:
git tag v0.2.0 && git push --tags— CI attaches the wheel and sdist to a GitHub Release.
Troubleshooting
plane-mcp: missing required env vars — the MCP host didn't pass the
PLANE_* variables. Re-run plane-mcp setup in the folder, or check the
env block in .mcp.json / .codex/config.toml.
no project registry found — create ~/.config/plane-mcp/projects.toml
(see Configure your projects).
The agent doesn't see the MCP server — make sure plane-mcp is on
PATH for the process that spawns the agent, and that the config file sits
in the folder you launched the agent from:
which plane-mcp # should print a path
plane-mcp mcp < /dev/null # should fail fast with a readable env-var error
License
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。