OpenChart Open-Source TradingView Alternative with AI Agents

OpenChart delivers AI‑enhanced charting without cloud quotas

OpenChart combines interactive financial charts with plug‑in AI agents, letting users run the full stack locally and avoid the subscription limits typical of commercial platforms.

Core capabilities are self‑contained and extensible

  • Unlimited charts and indicators – No software‑plan caps on the number of charts, technical studies, or price alerts.
  • Bring your own AI – Connect Claude, Gemini, Codex, or any compatible model via the user’s existing API keys.
  • Natural‑language alerts – Describe a market condition in plain text; OpenChart generates a Tea monitor that triggers an agent‑driven investigation when the condition fires.
  • AI‑generated explanations – Select a chart region and the attached agent researches the move, attaching source‑linked annotations directly to the bars.
  • Draw‑to‑alert workflow – Draw a line or zone; crossing it automatically launches an agent to analyze the breakout.
  • Tea scripting language – Write custom indicators and alert logic with full programming constructs (functions, loops, state, collections) that evaluate in real‑time as market data streams in.

Installation is a one‑click desktop download or a reproducible source build

The project provides a signed macOS (Apple Silicon) DMG for immediate use. For developers, the repository includes a reproducible build pipeline using Node 24, npm 11.11.1, Bun 1.4.2, and Just. The just install --frozen-lockfile command also compiles Tea, the domain‑specific language for market analysis.

"Download Desktop for macOS (Apple Silicon) … open the disk image, and install OpenChart." – OpenChart README

Running from source

git clone https://github.com/longsurf-ai/openchart.git
cd openchart
just install --frozen-lockfile   # installs deps and builds Tea
just desktop                      # launches the Electron‑based client

The same commands work on Linux and Windows after adjusting the binary download.

Pricing model isolates AI usage from the core app

OpenChart Desktop is free. AI usage follows the pricing of the connected model provider (Claude, Gemini, Codex, etc.). Optional cloud market‑data feeds are billed separately, but all charting, alerting, and scripting remain local and unrestricted.

Community response highlights privacy and integration strengths

  • Privacy‑first design – Users praised the local‑only storage of charts, research, and indicator code, noting it as a rare combination in the market‑analysis space.
  • Seamless Claude integration – Commenters highlighted the ease of plugging a Claude subscription into OpenChart compared with other tools.
  • Potential for mobile expansion – Some users expressed interest in iOS/Android clients to extend the workflow beyond desktop.

"AI agents + interactive charts is exactly the workflow I’ve been waiting for." – kirawen13 (HN comment) "Really like the design choice of keeping everything local (charts, research, indicator code). Privacy‑first + open source + agent‑native is a combination I haven't seen anyone else ship for markets." – aurielws (HN comment)

Open‑source ecosystem and acknowledgments

OpenChart builds on a rich stack of community projects:

  • Tea language – Inspired by TradingView’s Pine Script, enabling programmable market logic.
  • Effect runtime – Provides concurrency and error‑handling foundations.
  • assistant‑ui & AG‑UI – Power the agent conversation UI and event protocol.
  • shadcn/ui, Radix UI, Base UI – Supply reusable component libraries.
  • Robust‑orientation & kld‑intersections – Offer geometry algorithms adapted for chart annotations.

The repository lists additional credits, including Jan for visual design inspiration and OpenCode for conceptual guidance.

Contributing and governance

OpenChart welcomes contributions across code, indicator libraries, documentation, and user feedback. The maintainers recommend running just check before submitting pull requests and provide detailed development guides in the docs/ folder. Community channels include Discord, GitHub Issues, X (formerly Twitter), and a dedicated Reddit subreddit.

Security and licensing

Security vulnerabilities should be reported privately to security@longsurf.ai per the project's security policy. The code is released under the OpenChart License, a permissive Apache 2.0‑derived license with additional commercial usage conditions; commercial licensing inquiries are directed to legal@longsurf.ai.

Why OpenChart matters for traders and developers

OpenChart demonstrates that a fully local, open‑source trading‑chart platform can integrate modern LLM agents without imposing proprietary quotas. It offers a privacy‑preserving alternative to cloud‑centric services like TradingView, while giving power users the ability to script custom market logic and automate research through their preferred AI models.

Sources

Related

  • Project
  • Project
  • Project
  • Project
  • Dispatch