🦍 TradeApe
Open-source charting for the agent era. TradeApe pairs QuestDB-speed market data with KLineCharts and deterministic agentic tools, so every market read can be asked for, inspected, and extended.
Deterministic tools returned nearby support, reference VWAP context, and a rejected channel. The agent summarizes the evidence and keeps feed limits visible.
── Why TradeApe
Charts agents can actually use
TradeApe turns market questions into bounded tool calls for indicators, key levels, swing anchors, confluence, volume nodes, and structure before the assistant analyzes a chart.
Analysis with a paper trail
Every chart read keeps the feed, symbol mapping, schema, candle state, volume support, and exchange attribution visible, so the answer stays tied to the data it came from.
Hackable by design
Bring your own Anthropic key, run the app locally, inspect the prompts and tools, and fork the stack. No bundled model credits or hosted agent backend are hidden behind the product. Go bananas!
QuestDB is the engine room.
TradeApe builds on QuestDB's agent-native promise: standard SQL, time-series speed, and a shared source of truth for candles, volume, and market-structure queries. The agent asks for evidence; QuestDB supplies the time-series backbone.
KLineCharts, not screenshots
Interactive OHLCV charts, indicator panes, overlays, and layer controls give the agent a real canvas to mark up.
Market structure, bounded
Support, resistance, VWAP context, Fib anchors, ranges, channels, and wedges are generated with caps and caveats.
No magic chart calls
Nonparallel rails are rejected as channels, weak evidence stays weak, and trade recommendations stay out of scope.
Bring financial data
The current path is QuestDB-first, with provenance hooks built to make future feeds explicit instead of guessed.
Built for builders
The useful parts are exposed: prompts, deterministic tools, chart overlays, provenance objects, and validation scripts.
Open-source toolchain
AGPL TradeApe, QuestDB, and KLineCharts form a transparent base for agentic market-analysis experiments.
── Run the stack
── Writing queue
Why agentic charting needs deterministic tools
Language models are fluent at sounding like they're reading a chart. That fluency is the problem.
Read → infrastructureQuestDB as the market-data backbone
The queries do the work. The model explains what they found. A walkthrough of the time-series patterns behind every TradeApe tool call.
Read → designReplacing dashboards with inspectable agents
Dashboards answer the questions you had when you built them. Agents answer questions you didn't anticipate — but only if you can see through them.
Read →── FAQ
Does TradeApe include hosted AI access?
No. TradeApe is bring-your-own-key. You provide your own API key, and the app stores it locally in your browser. We strongly recommend a frontier model that supports tool calling. Consider Claude Sonnet/Opus 4.6 or better.
What makes QuestDB central here?
Agentic financial analysis needs fast, inspectable access to time-series data. QuestDB gives TradeApe a SQL-driven backbone for candles, latest state, sampled windows, and volume context without a proprietary data protocol in the middle. In short: It rocks.
How is TradeApe different from TradingView?
TradeApe is not trying to out-feature TradingView today. TradingView is a mature charting platform with broad market coverage, publishing, alerts, screeners, brokers, and community features.
TradeApe explores a different center of gravity: open-source, agent-first chart analysis where market reads are produced through inspectable tools, visible data provenance, and a hackable local stack.
Is the QuestDB demo feed enough for trading decisions?
No. The demo feed is historical Coinbase BTC-USDT data — useful for exploring the app, but not a substitute for live, venue-attributed market data. Use it to learn the tool, not to make trading calls.
Can I bring my own data?
Yes. Point the QUESTDB_URL env var at your own QuestDB instance
and set QUESTDB_TABLE, QUESTDB_EXCHANGE, and the
column variables to match your schema. See the
schema mapping guide
for the full setup.
Why does provenance matter?
Market analysis gets risky when software sounds certain about data it cannot verify. TradeApe surfaces provider, feed, symbol mapping, candle state, and volume or VWAP limits so each read stays calibrated.
Why agent-first instead of another dashboard?
Dashboards are useful when you already know exactly what to inspect. TradeApe is aimed at workflows where you ask a market question, the agent calls deterministic tools, marks the chart, and explains what the evidence supports.
Is this financial advice?
No. TradeApe is charting and analysis software. It can summarize deterministic tool output, but it should not be treated as trading advice or execution guidance.
How can I contribute?
Contributors are emphatically welcome. Open issues, send pull requests, improve docs, test the app with real workflows, or star the repo so more builders can find it. All of those help the project move.
Why AGPL?
TradeApe is meant to stay open when modified and offered as a hosted service. Self-hosted use is free and unrestricted, and service modifications should be shared back under the AGPL terms.