Chart Control
Let an AI assistant inspect TradingView symbols, timeframes, drawings, indicators, and replay state through a local bridge.
Planned MCP feature
Control charts. Run backtests. Journal the decision.
A local MCP bridge for TradingView workflows: AI can inspect charts, prepare tests, summarize edge, and hand every result back to your trading journal.
Claude / Codex controlled chart
+$401,830
5,944 tests81.1% WRSharpe 4.21
Top setup - BTC / momentum
x81.17
+$29,247
Probability lattice
Human approval gate
- Read-only by default
- Strategy must pass risk cap
- Live execution stays locked
Bridge status
Checking bridge status
Looking for a local MCP bridge session.
Checking
Disconnected
Start the local TradingView MCP bridge to enable chart context and backtest research.
Read-only
Chart inspection is available. Order placement remains disabled.
Testing
A research job can run with assumptions, time range, sample size, and failure modes captured.
Live-risk locked
This sprint does not allow order placement. User approval is required before research becomes a trade plan.
Last checked: Not checked yet
Read-only defaultOn
Order placementDisabled
Approval gateRequired
Audit trailPending backend
Backtests5,944
+401.8K simulated P&L
Win rate81.1%
Filtered by session setup
Edge score92.7
Signal confidence
Latency185ms
Local bridge target
Feature surface
Built for traders who want proof before conviction.
Backtest Runner
Convert a setup idea into a constrained test plan, then run historical checks before it ever touches a live account.
Risk Gate
Require max loss, session, instrument, and confidence checks before any strategy gets marked tradable.
Journal Loop
Send every test, rule, and result back into TradeMCP so the journal learns what actually improves your edge.
PM handoff
Codex owns the product page and UX. Claude owns the bridge internals.
This page frames the feature and creates the product contract. Backend transport, browser automation safety, broker permissions, and backtest correctness should stay in Claude's critical-integration lane.
- AI assistant prepares a chart or hypothesis
- Local MCP bridge validates permissions
- TradingView state and backtest data are inspected
- TradeMCP records the decision, rule, and result
Make AI useful without letting it trade unchecked.
The first release should be read-only, local-first, and approval gated. That gives traders faster research without turning the assistant into an unsupervised execution engine.
Share feedback