jev_vis_router
Python module that uses Jev to pick a primary and supporting chart from caller-supplied candidates, plus candlestick annotation planning.
Overview
JEV Visualization Router is a standalone Python module that chooses how to visualise an answer. Given a user question (for example, show monthly revenue trend) and caller-supplied chart component candidates with descriptions, it asks Jev to select a primary chart and an optional supporting chart, and returns the analysis angle, visual encoding, evidence strategy and marking strategy. A no-visual outcome is allowed, supporting charts are dropped if duplicated or below their confidence threshold (default 0.55), and a primary below 0.35 confidence, missing credentials, network failures or invalid candidates raise an error so the host can fall back.
A second feature plans candlestick chart annotations: Jev may only choose among candidate coordinates already validated for symbol, period, date and price range, or computed from supplied price data such as support and resistance or moving averages; it cannot invent future prices. The host does data fetching, schema validation and rendering. Tests use a mock SDK client, and documentation is in Chinese.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which candidate chart (or no-visual) best answers this question? | — |
| Choice | Which validated candidate annotations should be marked on this candlestick chart (or no-mark)? | — |
How it works
- Caller supplies question and chart candidates
- Jev selects primary and supporting charts with strategies
- Confidence thresholds drop weak picks or raise for fallback
- Host renders; for K-line, Jev picks only pre-validated annotation candidates
Strengths
- Model only selects among host-provided options
- Explicit thresholds and error-based fallback
- Tests run without API keys
Caveats
- Chinese documentation
- Does not fetch data or render charts itself
Getting started
python -m pip install -r requirements.txt, set TYPESAFE_API_KEY, then call JevVisualizationRouter().route(question=..., candidates=[...]).At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- Library / CLI
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
Standalone JEV visualization routing with chart selection, confidence…
wh000wh000/awesome-jev-live
Frequently asked questions
Can Jev invent new chart types?
No, it only picks among candidates the caller provides, or no-visual.
What happens on low confidence?
Weak supporting charts are dropped; a weak primary raises an error so the host can fall back.
Can it add made-up price levels to charts?
No, it can only choose annotation candidates validated against supplied data.
Related projects
Source for a shared emoji canvas where Jev picks the next emoji and its placement after every stroke and judges if the shape is unfinished.
ProjectproductionTypeScriptLocal China A-share technical analysis workbench where Jev assesses eight indicator dimensions and returns Buy, Hold or Sell with evidence.
ProjectbetaPython, TypeScriptLocal stock decision service that sends market data and news to Jev or a local model and returns a policy-checked buy/sell/hold.
ProjectbetaTypeScriptReact components that use Jev judgments to choose which component renders, how lists are ordered and whether affordances show.
ProjectbetaTypeScriptPrompt-first music studio where Jev makes every compositional choice, from plan to each bar pattern, rendered as editable MIDI.
ProjectbetaTypeScriptVercel Labs' generative UI framework with an experimental compose path where Jev picks which catalog components and actions to render.
ServicebetaTypeScript