Repository navigation
Yeti 1.0.0
Yeti is an MCP server that guides product design through five phases, from the problem to a verified scope, and hands execution over to Jira.
What is in 1.0.0
Design workflow
- Projects move through discovery, definition, prioritization, flows and verification. In
strictmode a phase advances only when it is complete; infreemode it advances and reports what was skipped. - Requirements with MoSCoW priorities, user stories linked to requirements, acceptance criteria you can tick, RICE scoring, Mermaid flows with Yes/No branches,
nextandends. - Verification requires every must to be done and covered by a user story.
Working with the backlog
- Edit, delete, search, archive and duplicate projects and items; project summary.
- Export to Markdown or CSV, optionally saved to a file; import requirements and stories from CSV.
Jira
push_to_jiracreates a Task per requirement and a Story per user story, with dry-run by default,limit, and no duplicates;updaterefreshes linked issues.sync_from_jiramarks items done when their issue is done and reopens what it completed when the issue goes back.
Dashboard
open_dashboard(ornpx yeti-mcp dashboard) serves a read-only page on 127.0.0.1: phases, what is missing, musts and their stories, criteria, RICE and flows. It refreshes every 10 seconds.
Storage
- Data lives in SQLite (
data/yeti.sqlite); adata/db.jsonfrom earlier versions is imported once.
Requirements
- Node.js 22.13 or later.
- Jira credentials only for the Jira tools (
JIRA_SITE,JIRA_EMAIL,JIRA_API_TOKEN).
Changes since 0.9.0
- New live dashboard and
open_dashboardtool. - The MCP server now reports its real version, read from
package.json.