Skip to content
AgentsOpen sourceOther

ATLAS

Self-improving trading agents whose prompts are rewritten when they lose, from General Intelligence Capital

By General Intelligence Capital

Live GitHub metrics

Live GitHub data · Oct 11, 2026
Stars
2,312
Stars · 7d
+8
Stars · 30d
+147
30d momentum
+6.8%
Forks
417
Open issues
4
Language
Python
Last push
19 days ago

No tagged releases. Repository created Mar 11, 2026.

307413518Sep 1Oct 10
Cumulative GitHub stars, Sep 1, 2026 to Oct 10, 2026. Source: OSSInsight star history.

What it is

ATLAS runs 25+ agents in four layers (macro, sector desks, investor personas and a decision layer with an adversarial risk officer) that debate markets daily. Every recommendation is scored against outcomes, the worst agent's prompt is rewritten and the change is kept only if performance improves, using git commits and reverts as the training loop in the style of Karpathy's autoresearch. The README states it runs live with real capital; treat that as the maintainers' claim.

Custom licence that is not OSI-approved; read it before reuse.

In the maintainers' words: “ATLAS by General Intelligence Capital — Self-improving AI trading agents using Karpathy-style autoresearch”

Markets

  • US equities

Tags

  • Self-improving
  • Multi-agent
  • Prompt evolution
  • Debate