Skip to main content

MCP server for Allstacks API - provides AI-ready interface to Allstacks metrics, projects, and analytics

Project description

Allstacks MCP Server

A comprehensive Model Context Protocol (MCP) server providing AI-ready access to the Allstacks API using HTTP Basic Authentication.

Overview

This MCP server exposes 208+ tools organized into 12 categories for comprehensive interaction with Allstacks:

Tool Categories

  1. Metrics & Analytics (18 tools): GMDTS data, Metrics V2, insight configs, population benchmarks, company metrics
  2. Service Items & Work Items (21 tools): Complete CRUD for work items, parent service items, property keys, estimation methods, notes, filter sets
  3. Users & Teams (23 tools): Full user management, invites, roles, team tags, personal access tokens, service users
  4. Organization & Projects (28 tools): Organizations, projects, settings, services, calendars, time periods, slots configuration
  5. Dashboards & Widgets (20 tools): Complete dashboard/widget CRUD, shared links, cloning, widget management
  6. Employee Analytics (9 tools): Employee metrics, cohorts, work items, timeline, summary, periods
  7. Forecasting & Planning (11 tools): V3 forecasts, velocity, scenarios, capacity planning, chart analysis
  8. Labels & Tagging (17 tools): Labels, label families, bulk operations, service item label assignment
  9. Alerts & Monitoring (16 tools): Alert rules, active alerts, notifications, subscriptions, preferences
  10. AI & Intelligence (18 tools): AI reports, Action AI code query, metric builder, pattern analysis, surveys, DX scores, AI tool usage
  11. Work Bundles (13 tools): Selectable work bundle management, forecasting, metrics, cloning
  12. Risk Management (14 tools): Risk definitions, project risks, assessment, trends, resolution

Project Structure

allstacks-mcp/
├── server.py                    # Main entry point - 208+ tools
├── allstacks_mcp/
│   ├── __init__.py
│   ├── client.py               # HTTP Basic Auth client
│   └── tools/                  # Tool modules by category
│       ├── __init__.py
│       ├── metrics.py          # 18 metrics tools
│       ├── service_items.py    # 21 service item tools
│       ├── users_teams.py      # 23 user/team tools
│       ├── org_projects.py     # 28 org/project tools
│       ├── dashboards.py       # 20 dashboard tools
│       ├── employee.py         # 9 employee analytics tools
│       ├── forecasting.py      # 11 forecasting tools
│       ├── labels.py           # 17 label management tools
│       ├── alerts.py           # 16 alert/monitoring tools
│       ├── ai_analytics.py     # 18 AI & analytics tools
│       ├── work_bundles.py     # 13 work bundle tools
│       └── risk_management.py  # 14 risk management tools
├── pyproject.toml
├── uv.lock
├── FILTERED_ENDPOINTS.md       # Endpoint verification document
├── TOOLS_MAPPING.md            # Complete API mapping
├── ENDPOINT_VERIFICATION.md    # Verification results
└── README.md

Installation

  1. Clone this repository
  2. Install dependencies using uv:
    uv sync
    

Authentication & Security

Required Credentials

The server uses HTTP Basic Authentication to connect to the Allstacks API. You'll need:

  • Username: Your Allstacks username or email address
  • Password: Your Allstacks password or API key
  • Base URL: (Optional) Default is https://api.allstacks.com/api/v1/

🔒 Security Best Practices

Credential Management

⚠️ IMPORTANT: Your credentials provide full access to your Allstacks account.

  1. Use Environment Variables (Recommended)

    export ALLSTACKS_USERNAME="your-username"
    export ALLSTACKS_PASSWORD="your-api-key"
    
  2. Use Allstacks API Keys instead of passwords when possible

    • Generate API keys from your Allstacks account settings
    • API keys can be revoked without changing your password
  3. Never commit credentials to version control

    • The MCP client config file may contain credentials
    • Keep config files in .gitignore
  4. Rotate credentials regularly for enhanced security

Process Security

Note: Command-line arguments (--username, --password) are visible in process lists. For production use:

  • Use environment variables
  • Use secure configuration files
  • Consider using a secrets management system

Data Access & Privacy

This MCP server acts as a pass-through to the Allstacks API:

  • ✅ Does not store or log your credentials
  • ✅ Does not cache API responses
  • ✅ Does not persist any data locally
  • ✅ Returns API data as-is without modification

AI Access: When used with AI assistants (e.g., Claude), the AI will have access to:

  • All data accessible via your Allstacks credentials
  • The same permissions as your Allstacks user account
  • Ability to create, modify, and delete resources (depending on your permissions)

Recommendation: Consider using a dedicated Allstacks account with limited permissions for AI access.

Usage

Running the Server

uv run server.py --username YOUR_USERNAME --password YOUR_PASSWORD

Command-line options:

  • --username or -u: Username for HTTP Basic authentication (required)
  • --password or -p: Password for HTTP Basic authentication (required)
  • --base-url or -b: Override the default API base URL (default: https://api.allstacks.com/api/v1/)

MCP Client Configuration

Add to your MCP client configuration (e.g., Claude Desktop's claude_desktop_config.json):

{
    "mcpServers": {
        "allstacks": {
            "command": "uv",
            "args": [
                "--directory",
                "/ABSOLUTE/PATH/TO/allstacks-mcp",
                "run",
                "server.py",
                "--username",
                "YOUR_USERNAME",
                "--password",
                "YOUR_PASSWORD"
            ]
        }
    }
}

Important Notes:

  • Replace /ABSOLUTE/PATH/TO/allstacks-mcp with the full path to this directory
  • Replace YOUR_USERNAME and YOUR_PASSWORD with your actual credentials
  • On macOS, uv is typically located at ~/.local/bin/uv - ensure it's in your PATH or use the full path
  • Claude Desktop doesn't inherit your shell's PATH, so using the full path to uv is more reliable

Alternative Configuration (Direct Python)

If you prefer to use Python directly from a virtual environment:

{
    "mcpServers": {
        "allstacks": {
            "command": "/ABSOLUTE/PATH/TO/allstacks-mcp/.venv/bin/python",
            "args": [
                "/ABSOLUTE/PATH/TO/allstacks-mcp/server.py",
                "--username",
                "YOUR_USERNAME",
                "--password",
                "YOUR_PASSWORD"
            ]
        }
    }
}

Features

Comprehensive API Coverage

All tools are verified against the official Allstacks OpenAPI specification:

  • ✅ All endpoints use correct paths and parameters
  • ✅ Detailed parameter descriptions from official API docs
  • ✅ Proper error handling and validation
  • ✅ HTTP Basic Auth support
  • ✅ Async/await for performance

Modular Architecture

Tools are organized into logical categories matching the Allstacks API structure:

  • Easy to navigate and understand
  • Clear separation of concerns
  • Consistent patterns across all tools
  • Well-documented with OpenAPI references

Rich Tool Descriptions

Every tool includes:

  • Purpose and use case
  • Required and optional parameters
  • Expected return formats
  • OpenAPI endpoint reference
  • Example usage patterns

Example Tool Usage

Get GMDTS Metrics Data

# Get Generated Metric Data Time Series (GMDTS)
get_gmdts_data(
    project_id=123,
    metric_type="Velocity",
    start_date=1704067200000,
    end_date=1706745600000
)

List Work Items with Filtering

# List service items with filtering
list_service_items(
    item_type="CARD",
    limit=100,
    offset=0
)

Create AI Report

# Generate AI analysis report
create_ai_report(
    org_id=456,
    report_type="project_health",
    project_id=123
)

Forecast Work Bundle

# Get Monte Carlo forecast for work bundle
get_work_bundle_forecast(
    project_id=123,
    bundle_id=789,
    confidence_level=80
)

API Compatibility

This server is built against the official Allstacks API v1 specification and includes:

  • Non-admin endpoints only (no destructive operations)
  • Full CRUD for resources where appropriate
  • Bulk operations for efficiency
  • Proper pagination support
  • Query parameter validation

Development

Adding New Tools

  1. Create or update a tool file in allstacks_mcp/tools/
  2. Follow the existing pattern for tool registration
  3. Include full docstrings with OpenAPI references
  4. Add the module to allstacks_mcp/tools/__init__.py
  5. Register it in server.py's register_all_tools() function

Testing

# Run the server in development mode
uv run server.py --username test_user --password test_pass --base-url https://api-dev.allstacks.com/api/v1/

License

MIT

Support

For issues or questions about the Allstacks API, contact the Allstacks team. For issues specific to this MCP server, please file an issue in the repository.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

allstacks_mcp-0.1.1.tar.gz (29.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

allstacks_mcp-0.1.1-py3-none-any.whl (36.4 kB view details)

Uploaded Python 3

File details

Details for the file allstacks_mcp-0.1.1.tar.gz.

File metadata

  • Download URL: allstacks_mcp-0.1.1.tar.gz
  • Upload date:
  • Size: 29.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.9.5

File hashes

Hashes for allstacks_mcp-0.1.1.tar.gz
Algorithm Hash digest
SHA256 5901b84abb0e43d4d0b228f8d4cba8ae9d693470e95b72265ee86bd81bf10d42
MD5 ad325c918056aea277c658470601f814
BLAKE2b-256 c2dccab23f547bd42966c7638e58bf6c37b08c5e062a661af80c1ca303053292

See more details on using hashes here.

File details

Details for the file allstacks_mcp-0.1.1-py3-none-any.whl.

File metadata

File hashes

Hashes for allstacks_mcp-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 3031e225a3ecf8dc691760397c55b5d7534e25df143cc1f58f893adff9c7ba78
MD5 51b9eb519f75aeceadda17cef6e31c5b
BLAKE2b-256 ffd0681bd47ffe49d86bea5034cb15f48b32e716118371f83feaccb1df769ec9

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page