Labsco
MCP SERVER

Taiga MCP Bridge

by talhaorak ↗

Drive a Taiga project from an assistant — projects, epics, user stories, tasks, issues and sprints — with a verbosity switch that keeps long list responses out of your context window.

Project & Task Management
Summary
The verbosity switch is the feature that makes bulk listing practical.

Taiga's API responses are wide, and a sprint's worth of user stories at full fidelity will eat a conversation. Asking for `minimal` on the listing call and `standard` only on the item you are about to update keeps that under control. Auto-authentication from environment variables removes the session bookkeeping for the common single-account case; the login and logout tools remain for juggling several.

What it is

A bridge between Taiga's project-management API and MCP. It covers the full Taiga object graph — projects, epics, user stories, tasks, issues and milestones — with create, read, update and delete on each, and exposes project data as `taiga://` resources as well as tools. Every tool takes a `verbosity` parameter, so a listing call can return five core fields instead of the complete API response.

What you get
  • Projects, epics and user stories created and listed against a Taiga instance — `create_project`, `create_epic`, `create_user_story`, `list_projects`, `list_user_stories`
  • Response size controlled per call: `minimal` returns id, ref, subject, status and project; `standard` adds the version field needed for updates; `full` returns the complete API response
  • Project data readable as resources rather than tool calls, for example `taiga://projects` and `taiga://projects/123/epics`
  • A session model for multi-account work: log in, check whether the session is still valid, and log out — `login`, `session_status`, `logout`
  • Allowlist-based parameter validation, so unexpected fields never reach the Taiga API, and credentials that are kept out of logs, error messages and stack traces
Requirements

A Taiga account. Python 3.12 or higher and the `uv` package manager; `uv pip install -e .` from a clone, or the prebuilt image `ghcr.io/tetra-2023/pytaiga-mcp:latest`. `TAIGA_API_URL` points at your instance and defaults to `http://localhost:9000`. Setting `TAIGA_USERNAME` and `TAIGA_PASSWORD` makes the server authenticate itself at startup, after which tool calls can omit the session id entirely; without them you call the login tool and pass the session id explicitly. `TAIGA_TRANSPORT` chooses `stdio` or `sse`, matching the `--sse` flag. The published package is `mcp-taiga-bridge` 2.0.1. The README describes this repository as a community fork with its own CI and ongoing maintenance.

Setup effort

One command plus a key — docker pull ghcr.io/tetra-2023/pytaiga-mcp:latest, then supply credentials