LangAlpha
Open-source financial research agent that runs code instead of eating your context window
LangAlpha is an open-source financial research agent built on Claude and LangChain that takes a fundamentally different approach to financial data: instead of injecting raw price series or filings into the context window, it writes and executes Python code in Daytona cloud sandboxes. Five years of daily OHLCV data for 500 tickers would consume tens of thousands of tokens as raw text — as executed code, it consumes almost none. Research compounds across sessions via persistent "workspaces" (e.g., "Q2 rebalance," "NVDA earnings deep-dive"). The agent ships 23 pre-built slash-command skills: DCF modeling, earnings transcript analysis, SEC filing review, macro overlays, and more. The Programmatic Tool Calling (PTC) architecture means the agent drafts, runs, and iterates on analysis code rather than retrieving static answers — closer to how an actual analyst thinks. The indie team open-sourced under Apache 2.0 and the HN Show HN thread highlights strong interest from quant developers and independent RIAs. The architecture pattern — code execution over data injection — is broadly applicable beyond finance and represents a meaningful contribution to the agent design space.
Panel Reviews
The Builder
Developer Perspective
“The PTC architecture is the right call — injecting raw financial time series into a context window was always the wrong abstraction. Persistent workspaces mean research actually accumulates instead of resetting each session. The 23 pre-built skills cover 80% of what a junior analyst does daily. Fork-worthy even if you don't use it as-is.”
The Skeptic
Reality Check
“Sandbox code execution on financial data raises real questions: how are API keys and brokerage credentials handled? Daytona sandbox cold starts could introduce latency in time-sensitive analysis. And 'AI-written Python for DCF models' needs robust human review — errors in financial models compound in bad ways.”
The Futurist
Big Picture
“The code-execution-over-data-injection pattern is going to become standard for data-heavy agent domains: genomics, legal discovery, supply chain analytics. LangAlpha is proving it in finance first, and the open-source architecture gives the community a reference implementation to fork for other verticals.”
The Creator
Content & Design
“For independent researchers and finance content creators, this is a serious productivity unlock — structured analysis that compounds over time instead of starting from scratch each session. The slash-command UX is clean and the output is already formatted for presentation.”
Community Sentiment
“PTC architecture and token efficiency vs. raw data injection”
“Security of sandbox execution with financial credentials”
“Apache 2.0 and whether it supports real-time data feeds”