Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery

Ao Qu, Han Zheng, Zijian Zhou, Yihao Yan, Yihong Tang, Shao Yong Ong, Fenglu Hong, Kaichen Zhou, Chonghe Jiang, Minwei Kong, Jiacheng Zhu, Xuan Jiang, Sirui Li, Cathy Wu, Bryan Kian Hsiang Low, Jinhua Zhao, Paul Pu Liang

Abstract

Large language model (LLM)-based evolution is a promising approach for open-ended discovery, where progress requires sustained search and knowledge accumulation. Existing methods still rely heavily on fixed heuristics and hard-coded exploration rules, which limit the autonomy of LLM agents. We present CORAL, the first framework for autonomous multi-agent evolution on open-ended problems. CORAL replaces rigid control with long-running agents that explore, reflect, and collaborate through shared persistent memory, asynchronous multi-agent execution, and heartbeat-based interventions. It also provides practical safeguards, including isolated workspaces, evaluator separation, resource management, and agent session and health management. Evaluated on diverse mathematical, algorithmic, and systems optimization tasks, CORAL sets new state-of-the-art results on 10 tasks, achieving 3-10 times higher improvement rates with far fewer evaluations than fixed evolutionary search baselines across tasks. On Anthropic's kernel engineering task, four co-evolving agents improve the best known score from 1363 to 1103 cycles. Mechanistic analyses further show how these gains arise from knowledge reuse and multi-agent exploration and communication. Together, these results suggest that greater agent autonomy and multi-agent evolution can substantially improve open-ended discovery. Code is available at https://github.com/Human-Agent-Society/CORAL.

CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery

Abstract

Large language model (LLM)-based evolution is a promising approach for open-ended discovery, where progress requires sustained search and knowledge accumulation. Existing methods still rely heavily on fixed heuristics and hard-coded exploration rules, which limit the autonomy of LLM agents. We present CORAL, the first framework for autonomous multi-agent evolution on open-ended problems. CORAL replaces rigid control with long-running agents that explore, reflect, and collaborate through shared persistent memory, asynchronous multi-agent execution, and heartbeat-based interventions. It also provides practical safeguards, including isolated workspaces, evaluator separation, resource management, and agent session and health management. Evaluated on diverse mathematical, algorithmic, and systems optimization tasks, CORAL sets new state-of-the-art results on 10 tasks, achieving 3-10 times higher improvement rates with far fewer evaluations than fixed evolutionary search baselines across tasks. On Anthropic's kernel engineering task, four co-evolving agents improve the best known score from 1363 to 1103 cycles. Mechanistic analyses further show how these gains arise from knowledge reuse and multi-agent exploration and communication. Together, these results suggest that greater agent autonomy and multi-agent evolution can substantially improve open-ended discovery. Code is available at https://github.com/Human-Agent-Society/CORAL.

Paper Structure

This paper contains 66 sections, 6 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Coral: A Framework for Autonomous Multi-Agent Evolution
  4. Preliminaries: Problem Formulation for Open-Ended Discovery
  5. From Fixed Search to Autonomous Multi-Agent Evolution
  6. Core Mechanisms of CORAL
  7. Experiments
  8. Experimental Setup
  9. Autonomous Evolution Outperforms Fixed Evolutionary Search
  10. Multi-Agent Evolution Extends the Search Frontier
  11. Analysis
  12. Why Autonomous Evolution Works
  13. Why Multi-Agent Organization Helps
  14. Ablations
  15. Conclusion
  16. ...and 51 more sections

Figures (6)

  • Figure 1: Comparison of three paradigms for LLM-based open-ended discovery.
  • Figure 2: Overview of the CORAL framework. Autonomous agents operate in isolated worktrees, iteratively propose and evaluate candidate solutions, and accumulate shared persistent memory (attempts, notes, skills) through a hub. Heartbeat-driven periodic reflections help agents consolidate discoveries and reorient search over long horizons.
  • Figure 3: Polyominoes packing: single-attempt baseline (left, 56.0%) vs. CORAL (right, 89.4%) uses Claude Opus 4.6 via Claude Code with web search access. The CORAL solution surpasses the previous best known score of 87%.
  • Figure 4: Architecture of CORAL. The system is organized into six modules: Configuration parses YAML task definitions; the Agent System manages agent lifecycles and heartbeat-driven interventions; the Grader Hierarchy provides a pluggable evaluation interface; Workspace Setup creates isolated per-agent worktrees with symlinks to shared state; the Hub stores shared persistent memory (attempts, notes, skills); and Core Types define the data model. Arrows indicate primary data flow: configuration is consumed by both the agent system and grader loader; workspace setup creates symlinks into the hub; and graders return ScoreBundle objects defined in core types.
  • Figure 5: CORAL user interface. The interface supports both trajectory monitoring and knowledge inspection during experiments.
  • ...and 1 more figures