Two parts do the heavy lifting. Batch creation takes up to twenty items at once and reports each outcome separately, so a sprint's worth of stories is one call instead of twenty. And the query syntax is precise enough to be worth learning — run it past the help and validation tools first rather than guessing at the operators.
An MCP server over the Taiga API, authenticated with your own account, covering the whole board rather than just the ticket list.
- Projects and sprints: list projects and read one, list milestones, create one, read its progress statistics, and pull the issues inside it
- User stories: list, read, create, assign to a sprint and update status
- Tasks created against a user story
- Issues end to end: list, read, create, add to a sprint, assign, and update status
- Batch creation for issues, user stories and tasks — up to twenty at a time, with each item's success or failure reported separately so one bad row does not sink the rest
- An SQL-like query language: field:operator:value combined with AND, OR and NOT, fuzzy and wildcard text matching, date ranges, and ORDER BY and LIMIT — with a help tool that explains the syntax and a validator that checks a query before you run it
- Comments on any work item: add, list, edit and delete
- Attachments uploaded from a file path or as base64, then listed, downloaded and deleted
- Epics for the layer above stories: create, list, read and update them, and link or unlink user stories
- A wiki: create, list, read, update, delete and watch pages
A Taiga account with API access, and Node.js v14 or higher. Three environment variables carry the connection: TAIGA_API_URL (https://api.taiga.io/api/v1 for the hosted service, or your own instance's /api/v1 endpoint), TAIGA_USERNAME and TAIGA_PASSWORD. Run it over stdio with npx taiga-mcp-server, or as a Docker container reading the same variables from an env file or -e flags.
One command plus a key — npx taiga-mcp-server, then supply credentials