Trello MCP Server
Enables human-in-the-loop AI development where Trello tickets drive analysis, implementation, commits, and board sync in Cursor.
README
MCP Dev Loop: Trello → Cursor → GitHub
Human-in-the-loop AI development — Trello tickets drive analysis, implementation, commits, and board sync in Cursor.
❌ Not autonomous AI — the agent suggests, you approve, the agent executes.
✅ Human-in-the-loop — no surprise commits, pushes, or Trello updates.
Clone this repo, connect Trello via MCP, and run a repeatable dev loop your whole team can use.
Architecture
┌──────────────┐ ┌───────────────────┐ ┌─────────────────┐
│ Trello │ ──► │ Trello MCP Server │ ──► │ Cursor Agent │
│ (Tickets) │ │ (this repo) │ │ (you control) │
└──────────────┘ └───────────────────┘ └────────┬────────┘
│
┌───────────────────────────────┼───────────────────────────────┐
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Codebase │ ─────────────► │ GitHub │ ─────────────► │ Trello │
│ Node/React… │ commit/PR │ commits/PRs │ update/sync │ comments │
└─────────────┘ └─────────────┘ └─────────────┘
Repo structure
| Path | Purpose |
|---|---|
trello-mcp-server/ |
Core engine — Trello API + MCP tools |
cursor-rules/ |
Agent rules, skill, approval hooks |
workflows/ |
Command system (analyze ticket, implement ticket, …) |
examples/ |
End-to-end session walkthroughs |
docs/ |
Setup, MCP, safety model, example prompts |
bin/ |
install.sh, sync-global-cursor.sh |
.cursor/ |
Ready-to-use Cursor config (MCP + rules + hooks) |
Quick start
git clone https://github.com/Mih-Nig-Afe/MCP-Dev-Loop-Trello-Cursor-GitHub.git
cd MCP-Dev-Loop-Trello-Cursor-GitHub
./bin/install.sh
./bin/sync-global-cursor.sh
- Edit
.envwith Trello API credentials npm run test-api— verify connection + full card extraction- Cursor Settings → MCP → confirm
trellois connected (Refresh if needed) - In chat:
analyze my assigned tasks
Full guide: docs/setup.md
Daily workflow
| Step | You say | Agent does |
|---|---|---|
| 1. Pull | analyze my assigned tasks |
Lists your Trello queue |
| 2. Plan | analyze ticket <id> |
Full card extraction → PLAN (no code) |
| 3. Approve | proceed |
— |
| 4. Build | implement ticket <id> |
Edits codebase |
| 5. Commit prep | prepare commit for ticket <id> |
Shows diff + draft message |
| 6. Commit | commit this change |
Creates commit (you approve via hook) |
| 7. Sync | update trello ticket <id> |
Comment + attach commit URL |
Command reference: workflows/README.md
Example sessions: examples/
Commands
analyze my assigned tasks
analyze ticket <cardId>
implement ticket <cardId>
fix issue in ticket <cardId>
prepare commit for ticket <cardId>
commit this change
update trello ticket <cardId>
mark in progress
mark done
Slash aliases: /analyze-ticket, /implement-ticket, /prepare-commit, /update-trello
Safety (enforced)
| Action | Protection |
|---|---|
git commit |
Cursor hook → asks your approval |
git push |
Hook → asks approval (main/master extra gate) |
| Trello comment / move | Hook → blocked unless you said update trello or mark done |
| Agent rules | Analyze phase = no code; commit/sync only on explicit command |
Details: docs/safety-model.md
MCP tools
get_my_cards · get_card (full extraction) · get_card_comments · add_comment · update_card · move_card · attach_commit · get_boards · get_board_cards · get_board_lists · mark_in_progress · mark_done
get_card returns description, all comments, attachments, checklists, custom fields, stickers, activity, and list/board context in one call.
Use in another project
./bin/sync-global-cursor.sh— registers MCP globally + copies skill- Copy
cursor-rules/into your app's.cursor/ - Keep
.envin this repo (global MCPcwdpoints here)
Guide: docs/cursor-setup.md
Stay up to date
git pull
./bin/sync-global-cursor.sh
npm run install:mcp # if dependencies changed
Then Refresh MCP in Cursor Settings.
Roadmap
- Auto-task prioritization
- Codebase memory layer
- PR auto-review
- Test runner integration before commit
- Slack/Telegram notifications
License
MIT — see 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。