Skip to content

Give AxonX a Star on GitHub! ↗

AGENT-NATIVE QUANT HARNESS

Financial quant research.
Connected. Traceable.

An agent-native harness for financial quantitative research. Organize algorithms as plugins, preserve evidence as Tasks, and connect Studio, CLI / MCP, and agents.

PLUGIN JOBTASKRESULT
From data to evidenceAXONX / RESEARCH WORKSPACE
↗

Latest Updates

  • AxonX 0.1.0 released

    Plugin-based quantitative research Tasks, execution tracking, task lineage, and shared CLI / MCP / Studio access. Documentation ↗

  • Connect your Agent with SKILL.md + CLI

    Load the Skill into Codex, Claude Code, or another Agent to discover Task contracts, develop plugins, submit research, and inspect results. SKILL.md ↗

  • AxonX Studio available

    Browse tasks and artifacts, inspect training curves and backtests, and compare strategies. The demo uses simulated data and execution. Try Playground ↗

  • Alpha158 Enhanced developed with Skill + CLI

    26 new features. In the 2025-01-01–2026-09-30 confirmation period, Top10 net annualized return rose from −5.74% to 28.21%, and Top20 from −3.24% to 24.93%. Explore the results ↗

01 / GET STARTED

Run your first task

Submit, follow, and inspect a built-in demo task.

↗

02 / RESEARCH

Build a research workflow

Connect data, features, training, predictions, and backtests.

↗

03 / AGENT

Connect your agent

Choose an external Skill / CLI / MCP integration or built-in Studio sessions.

↗

01 / AXONX STUDIO

See your research unfold.

Explore Studio ↗

From task execution to result interpretation, inspect your research in one workspace.

02 / HARNESS CORE

Flexible research. Grounded execution.

03 / CONTINUE YOUR RESEARCH

Continue by goal.

Documentation map ↗

04 / AGENT RESEARCH CASE

From plugin development to confirmation.

Experiment design ↗

Using Skill + CLI, Codex extended Alpha158 as a separate plugin with 26 new features, then completed ablations and independent confirmation. Confirmation-period Top10 net annualized return rose from −5.74% to 28.21%, and Top20 from −3.24% to 24.93%.

BUILD YOUR RESEARCH LOOP

Start with one Task.

The built-in demo needs no external data or model credentials.

Agent-native quant research.