Skip to content

current-status.md

Latest commit

 

History

History
180 lines (124 loc) · 4.78 KB

current-status.md

File metadata and controls

180 lines (124 loc) · 4.78 KB

Current Status

This document describes the repo as of March 18, 2026.

Deployment target

This app is designed for trusted single-user local desktop use only.

It is acceptable for:

  • one local desktop user
  • localhost-only backend access
  • trusted user-authored Pine and Python strategy code

It is not intended for:

  • networked deployment
  • multi-user access
  • untrusted code execution

See Production Readiness for the final position.

Finished

Desktop product

  • Windows desktop installer
  • Electron startup health polling
  • packaged FastAPI backend + embedded frontend SPA
  • first-run seeded demo dataset and demo replay run
  • direct Workspace opening on first launch

Data import and storage

  • Excel preview
  • CSV preview
  • inferred column mapping
  • dataset save to normalized CSV
  • saved dataset listing
  • atomic JSON writes with per-file locking for local artifact storage
  • isolated test data root for pytest

Run orchestration

  • replay run creation
  • dataset-backed incremental live playback
  • run persistence
  • run listing and retrieval
  • run progress WebSocket
  • live run cancellation and cleanup
  • failed-state handling for live worker crashes

Python execution

  • local Python execution contract
  • indicator extraction from numeric and boolean outputs
  • simple trade extraction from long_condition
  • wall-clock timeout around strategy execution
  • trusted-local hardening for the local runtime model

Pine workflow

  • PineTS-based local Pine execution in the frontend
  • manual bridge artifact upload and persistence
  • bridge artifact selection during runs

Comparison and library

  • indicator series comparison
  • trade comparison
  • mismatch classification and first mismatch output
  • Alignment page
  • built-in indicator library
  • custom save-to-library flow with persisted series_names

Verification and release tooling

  • backend automated tests
  • frontend production build
  • Pine built-in parity certification
  • Python built-in parity certification
  • combined Pine + Python parity report
  • Playwright smoke coverage for the main routes
  • desktop installer smoke checklist
  • release checklist

Ollama

  • local model discovery
  • chat-capable filtering
  • default model selection
  • prompt shaping and sanitization
  • fallback behavior for missing model, timeout, offline, and cleaned-empty response
  • machine-agnostic Ollama path resolution via env, PATH, and fallback discovery

Partial

Charts

  • candles render
  • price-like overlays render
  • oscillator-style series can be classified into a sub-pane
  • charting is still not a full multi-pane technical-analysis workspace

Pine truth workflow

  • local PineTS execution works for the supported Pine surface
  • exact TradingView truth still relies on manual bridge artifacts for some scripts
  • automated TradingView bridge does not exist

Live mode

  • incremental playback from a saved dataset works
  • provider-backed real market streaming is not implemented

Approval workflow

  • permission grants and TTL behavior work
  • UI can toggle approval entries
  • patch generation/application from the UI is still incomplete

Planned or intentionally not built

  • network auth or multi-user tenancy
  • production-grade untrusted-code isolation
  • provider-backed Polygon fetch flow
  • automated TradingView bridge worker
  • local Pine subset runtime separate from PineTS
  • broker or exchange-connected trading
  • token streaming in chat
  • CI-managed smoke orchestration without manual server startup

Known limitations

Trusted-local security model

The Python runtime is acceptable only for the trusted-local desktop target. It must not be treated as a production-grade sandbox for untrusted users.

File import scope

Local file import is intended for supported spreadsheet and CSV files only.

Active storage model

The running app currently uses JSON and CSV artifact storage under data/ or %APPDATA%. Older plans that mention SQLite or DuckDB as active storage are historical.

Live mode naming

live currently means replay-style bar-by-bar progression from a saved dataset. It is not an external market feed.

Provider gap

polygon remains visible in shared contracts and some UI paths, but active provider-backed fetching is not implemented.

Pine truth gap

Some Pine strategies still need manual bridge artifacts for exact parity checking.

Recommended next work

The forward-looking backlog is tracked in:

The short version:

  1. Canonical certification on a real dataset such as SBIN
  2. CI-friendly smoke entrypoint
  3. Auto-update and installer versioning
  4. Real provider-backed live data
  5. PineTS performance work on large datasets

Release summary

  • Trusted local desktop use: yes
  • Distribution-ready v1: yes
  • Networked or multi-user production system: no