tpm-mcp
An AI-powered Technical Project Manager that enables Claude to track projects, features, and tasks locally using SQLite. Break down complex features, track progress and blockers, and manage hierarchical project structures through natural conversation.
README
<p align="center"> <img src="assets/logo.svg" alt="tpm-mcp logo" width="120" /> </p>
<h1 align="center">tpm-mcp</h1>
<p align="center"> <strong>Your AI-powered Technical Project Manager in a box</strong> </p>
<p align="center"> <a href="#installation"><img src="https://img.shields.io/badge/uv-compatible-blue?style=flat-square" alt="uv compatible"></a> <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-green?style=flat-square" alt="MIT License"></a> <a href="#"><img src="https://img.shields.io/badge/python-3.10+-blue?style=flat-square" alt="Python 3.10+"></a> <a href="#"><img src="https://img.shields.io/badge/MCP-server-purple?style=flat-square" alt="MCP Server"></a> </p>
<p align="center"> <em>Have standups with Claude. Track features. Break down tasks. Ship faster.</em> </p>
What is this?
tpm-mcp is a local MCP (Model Context Protocol) server that gives Claude the ability to track your projects, features, and tasks. Think of it as having a dedicated TPM who:
- Remembers what you're working on across sessions
- Breaks down complex features into manageable tasks
- Tracks progress and blockers
- Gives you instant status updates
All data stays local in a fast SQLite database. No cloud. No subscriptions. Just you and your AI TPM.
Demo
You: :TPM: what's in progress?
Claude: Here's your current roadmap:
# Roadmap Summary
**Stats**: 3/8 tickets, 12/25 tasks (48% complete)
## ExampleOrg
### backend-api
Tickets: 3/5 done
- [~] **FEAT-003**: Payment Integration (high)
Tasks: 2/4
- [~] TASK-003-1: Stripe webhook handler
- [ ] TASK-003-2: Invoice generation
- [ ] **FEAT-004**: Email Notifications
Tasks: 0/3
Features
| Fast | Sub-millisecond queries with SQLite + WAL mode |
| Local | All data in ~/.local/share/tpm-mcp/tpm.db |
| Natural | Just talk - "I finished the auth feature" |
| Hierarchical | Orgs → Projects → Tickets → Tasks |
| Rich Metadata | Priority, tags, assignees, complexity, notes |
| Export/Import | Full JSON export/import for backup and migration |
Installation
1. Clone & Install
git clone https://github.com/yourusername/tpm-mcp.git
cd tpm-mcp
uv venv && uv pip install -e .
2. Add to Claude Code
claude mcp add tpm --scope user -- uv run --directory /path/to/tpm-mcp tpm-mcp
<details> <summary>Or manually edit settings.json</summary>
Add to ~/.claude/settings.json:
{
"mcpServers": {
"tpm": {
"command": "uv",
"args": ["run", "--directory", "/path/to/tpm-mcp", "tpm-mcp"]
}
}
}
</details>
3. (Optional) Enable TPM Agent Behavior
For the full experience, add the TPM agent instructions:
# Symlink to your project's .claude folder
mkdir -p .claude
ln -s /path/to/tpm-mcp/agents/tpm.md .claude/tpm.md
Then in your CLAUDE.md:
See .claude/tpm.md for project tracking workflow.
Usage
Quick Commands
| Say this... | Claude will... |
|---|---|
:TPM: status |
Show full roadmap |
:TPM: add auth feature |
Create a new ticket |
I finished the payment integration |
Mark ticket done |
Break down the API refactor |
Create subtasks |
What's blocking us? |
Show blocked items |
Workflow Example
You: :TPM: Add user authentication to the backend project
Claude: Created ticket FEAT-005: "User Authentication" in backend-api
Status: backlog, Priority: medium
You: Make it high priority and break it down
Claude: Updated FEAT-005 to high priority. Created tasks:
- TASK-005-1: Design auth schema
- TASK-005-2: Implement JWT middleware
- TASK-005-3: Add login/logout endpoints
- TASK-005-4: Write auth tests
You: Starting on the JWT middleware
Claude: Marked TASK-005-2 as in-progress.
FEAT-005 is now in-progress.
You: Done with middleware, tests pass
Claude: Marked TASK-005-2 as done (1/4 tasks complete)
Tools Reference
<details> <summary><strong>View Tools</strong></summary>
| Tool | Description |
|---|---|
roadmap_view |
Full roadmap with all orgs, projects, tickets, tasks |
ticket_get |
Detailed view of one ticket with its tasks |
info |
Server stats and database info |
</details>
<details> <summary><strong>Ticket Tools</strong></summary>
| Tool | Description |
|---|---|
ticket_create |
Create ticket with title, description, priority, tags |
ticket_update |
Update status, priority, assignees, etc. |
ticket_list |
List tickets, filter by project or status |
</details>
<details> <summary><strong>Task Tools</strong></summary>
| Tool | Description |
|---|---|
task_create |
Create task under a ticket |
task_update |
Update task status, details, complexity |
task_list |
List tasks, filter by ticket or status |
</details>
<details> <summary><strong>Organization Tools</strong></summary>
| Tool | Description |
|---|---|
org_create |
Create organization |
org_list |
List all organizations |
project_create |
Create project under org |
project_list |
List projects |
note_add |
Add note to any entity |
</details>
Data Model
Organization
└── Project
└── Ticket (feature/epic/bug)
├── Tasks
└── Notes
Ticket Statuses: backlog → planned → in-progress → done | blocked
Task Statuses: pending → in-progress → done | blocked
Priorities: critical > high > medium > low
Export & Import
Export and import your project data to JSON for backup, restore, or interoperability with other tools.
Export Data
Export all your project data to JSON:
# Export to stdout
uv run tpm-json-export
# Export to file
uv run tpm-json-export -o backup.json
# Export from custom database
uv run tpm-json-export --db-path /path/to/custom.db -o backup.json
Import Data
Import data from a JSON export file to recreate your database or restore from backup:
# Validate JSON file without importing
uv run tpm-json-import --dry-run backup.json
# Import into default database
uv run tpm-json-import backup.json
# Import into custom database
uv run tpm-json-import --db-path /path/to/db.db backup.json
# Clear existing data and import
uv run tpm-json-import --clear backup.json
The JSON format includes all data: organizations, projects, tickets, tasks, notes, and task dependencies. Use cases:
- Backup & Restore: Export your data regularly, restore if database gets corrupted
- Recreate Database: Start fresh by importing from a previous export
- Export to Other Tools: Use the JSON format to migrate to other project management tools
- Sync Between Machines: Export on one machine, import on another
PDF Status Reports
Generate beautiful, shareable PDF reports from your project data.
<p align="center"> <img src="assets/sample-report.png" alt="Sample PDF Report" width="500" /> </p>
<details> <summary><strong>Setup & Usage</strong></summary>
Using the Report Skill
The tpm-report skill teaches Claude how to generate professional status reports. To enable it:
# Copy the skill to your project
cp -r /path/to/tpm-mcp/skills/tpm-report .claude/skills/
Then ask Claude:
You: Generate a project status report
Claude: [Fetches roadmap data, generates styled HTML, converts to PDF]
PDF report saved to: Project-Status-2025-12-02.pdf
Prerequisites
- tpm-mcp: This MCP server (for
roadmap_viewdata) - Playwright MCP: For automatic HTML → PDF conversion
Installing Playwright MCP
# Add Playwright MCP (headless mode recommended for PDF generation)
claude mcp add playwright-headless --scope user -- npx @playwright/mcp@latest --headless
# Or with isolated mode (separate browser profile)
claude mcp add playwright-headless --scope user -- npx @playwright/mcp@latest --isolated --headless
<details> <summary>Or manually edit settings.json</summary>
Add to ~/.claude/settings.json:
{
"mcpServers": {
"playwright-headless": {
"command": "npx",
"args": ["@playwright/mcp@latest", "--headless"]
}
}
}
</details>
Without Playwright, Claude will generate an HTML file you can manually print to PDF.
What's in the Report?
| Section | Contents |
|---|---|
| Progress Overview | Visual progress bars for tickets and tasks |
| Project Breakdown | Completed, in-progress, and backlog items per project |
| Key Milestones | Major achievements and current focus |
| Blockers & Risks | Items requiring attention |
Manual Report Generation
If you prefer manual control, use roadmap_view with JSON format:
You: :TPM: show me the roadmap as JSON
Claude: [Returns structured JSON data]
Then use your preferred tool to format the output.
</details>
Migration from Legacy Trackers
Coming from a legacy JSON-based tracker with a different format?
uv run tpm-migrate /path/to/old-tracker
This tool converts from older JSON tracker formats. For standard export/import, use tpm-json-export and tpm-json-import above.
Development
# Install with dev dependencies
uv pip install -e ".[dev]"
# Run tests
uv run pytest tests/ -v
Why Local?
- Privacy: Your project data never leaves your machine
- Speed: SQLite is incredibly fast for this use case
- Reliability: No API rate limits, no outages, works offline
- Simplicity: One database file, easy to backup or sync
License
MIT - do whatever you want with it.
Citation
@software{Bhatia_TPM-MCP_A_Local_2025,
author = {Bhatia, Urjit Singh},
license = {MIT},
title = {{TPM-MCP: A Local Technical Product Manager MCP Server}},
url = {https://github.com/urjitbhatia/tpm-mcp},
year = {2025}
}
<p align="center"> <sub>Built with vibes ✨</sub> </p>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Neon Database
MCP server for interacting with Neon Management API and databases