Atlasmind is an AI-powered Jira project visualiser that lets you query your project in plain English and instantly see issue types as interactive hierarchy maps, tables and auto-generated charts. Try it live: atlasmind.de
- Natural‑language queries → JQL generated behind the scenes
- Hierarchy view (epic → story → subtask), tables, and charts in sync
- Demo mode with public Jira data, live mode for your own instance
| Tool | Purpose |
|---|---|
| Node.js ≥ 18 | Svelte dev server |
| uv | Python package manager |
| AtlasMind | Backend AI server (only needed for Live mode) |
Build the UI once, then start the bridge server. The bridge serves the built UI and proxies API calls to AtlasMind.
# 1. Build the Svelte app
cd jira-viz
npm install
npm run build
# 2. Start the AtlasMind backend (port 8000) — separate repo
cd ../AtlasMind
uv run python main.py
| URL | Mode |
|---|---|
http://localhost:8001/live |
Live — queries forwarded to AtlasMind → Jira |
http://localhost:8001/api/health |
Health check |
cd jira-viz
npm install
npm run dev # http://localhost:5173Also start the bridge in a second terminal:
uv run main.pyUse the Demo / Live toggle in the chat panel header to switch modes.
All settings are optional — defaults work for local development.
| Env var | Default | Purpose |
|---|---|---|
ATLASMIND_URL |
http://localhost:8000 |
AtlasMind backend URL |
JIRA_PAT |
(none) | Fallback PAT for Docker/local (UI entry takes precedence) |
JIRA_URL |
(none) | Jira server base URL forwarded to AtlasMind |
ALLOWED_ORIGINS |
Vite dev server | Comma-separated CORS origins for production |
HOST |
127.0.0.1 |
Bridge server bind address (set to 0.0.0.0 in containers) |
PORT |
8001 |
Bridge server port |
AGGREGATION |
true |
Set to false or 0 to disable the pandas aggregation pipeline |
CLI flags (alternative to env vars):
uv run main.py --external <IP> # connect to AtlasMind at a remote IP
uv run main.py --no-aggregation # disable pandas aggregation pipeline| Method | Path | Purpose |
|---|---|---|
POST |
/api/query |
Send a natural-language query; forwarded to AtlasMind |
POST |
/api/event |
Send a control event (cancel, heartbeat) to AtlasMind |
POST |
/api/aggregate |
Pre-aggregate issues server-side via pandas |
GET |
/api/meta |
Return model name and LLM timeout from AtlasMind |
GET |
/api/health |
Health check — reports whether AtlasMind is reachable |
In Live mode, a PAT (Personal Access Token) is required to query a private Jira instance.
- Open the app at atlasmind.de
The bridge forwards it as X-Jira-Token to AtlasMind, which uses it to authenticate against the Jira REST API.
For Docker/server deployments, set JIRA_PAT as an environment variable instead — the UI prompt is skipped automatically.
Click Ask AI (top-right) to open the chat panel.
- Live mode — sends the query to AtlasMind, which generates JQL and queries Jira; results auto-render as charts
The mode toggle shows a pulsing green dot when the AtlasMind backend is reachable, or dims to grey when offline.
Example queries:
- list all open bugs in project KAFKA
- who are the top 10 reporters and why types of issues do they log in project hive
- show me all the top burning topics in project hadoop, flink
When a Live query returns issues, the view switches automatically to the AI Chart tab. Additional auto-generated chart tabs (by status, priority, assignee, type) are shown alongside the AI-specified chart.
Each assistant reply shows:
- Token usage — total LLM tokens consumed by the query (displayed below the message bubble)
- Elapsed time — wall-clock time from send to response
While a query is in-flight the Send button turns into a Stop button. Clicking it sends a cancel event to AtlasMind via /api/event, which signals the backend to abort the LLM call early.
cd jira-viz
npm run build
# Output in jira-viz/dist/The bridge server (main.py) serves jira-viz/dist/ directly — no separate reverse proxy is needed.
For cloud deployment, set HOST=0.0.0.0, ALLOWED_ORIGINS=https://yourdomain.com, and JIRA_PAT (or rely on per-request PAT entry from the UI). HTTPS via a load balancer is recommended to protect the PAT in transit.
Designed to be easily deployed inside a private network and connect to any Jira Cloud or Data Center instance using a personal access token.