Allstacks MCP Server
Early access — not yet GA. The Allstacks MCP server is in active development. This section is for Allstacks team members and early-access design partners; it opens to customers when the beta launches. Currently documents allstacks-mcp v0.1.6.
The Allstacks MCP server lets AI assistants like Claude work directly with your Allstacks data. Connect it once and you can ask questions about your teams, delivery risks, and engineering metrics in plain language — from Claude Desktop, Claude Code, Cursor, or any MCP-compatible client.
What you can do
Explore your org — list workspaces/teams, contributors, roles, and connected tools (Jira, GitHub, Slack, and more)
See what's at risk — the risk rules configured for a workspace and how many items are currently flagged on each (stuck issues, aging PRs, oversized commits)
Build metrics with plain language — describe the chart you want ("investment hours by initiative, trailing 90 days") and the AI metric builder turns it into a Metrics V2 query and returns the data
Start from your org's metric templates — ready-made Metrics V2 templates covering velocity, cycle time, DORA, AI tool adoption, investment hours, and more
In this section
Install & Setup — install uv, create a Personal Access Token, and add the server to your client
Prompt Library & Recipes — example prompts grouped by what you're trying to do
Troubleshooting & FAQ — common setup and connection issues
Security & Data Notes — how the server handles credentials, scope, and AI token usage
Related
Product Studio — Allstacks' AI requirements workbench (the MCP is the agent-ready data plane alongside it)
AI Metrics — AI tool adoption and impact reporting in the Allstacks platform
Source of truth for setup details: the repo README.