SwarmResearch: Orchestrating Coding Agents for Open-Ended Discovery
Quick Answer
SwarmResearch introduces an orchestrator-subagent model that enhances coding agents' discovery capabilities for open-ended problems.
Quick Take
SwarmResearch introduces an orchestrator-subagent model that enhances coding agents' discovery capabilities for open-ended problems. By utilizing a Shepherd Agent to guide multiple Search Agents, it outperforms state-of-the-art techniques on 13 out of 15 optimization tasks, demonstrating superior exploration and adaptability in parallelism.
Key Points
- SwarmResearch uses a Shepherd Agent to guide multiple Search Agents.
- It outperformed state-of-the-art LLM-guided evolution on 13 out of 15 tasks.
- The model adapts parallelism at different search depths for improved performance.
- Long-running coding agents often converge on suboptimal solutions.
- The approach enhances exploration by maintaining global context.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Long-running coding agents such as autoresearch can persistently discover optimizations for open-ended problems. However, they tend to converge onto a single high-level approach, then proceed with low-level edits while missing other superior approaches to the problem. We hypothesize two harness-level design choices contribute to this behavior: accumulating context in a single long-running agent and only exposing a single program state to edit. We introduce SwarmResearch, an orchestrator-subagent harness in which a Shepherd Agent uses global context to steer a population of Search Agents, each operating with local context in their respective git branch. On open-ended optimization tasks, SwarmResearch discovers better or comparable solutions to state-of-the-art LLM-guided evolution and multi-agent techniques on 13/15 tasks, driven by higher-level exploration. Compared with fixed scaling of serial and parallel agents, SwarmResearch's orchestrator-guided scaling discovers better-performing solutions by adapting parallelism at different search depths.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.02807 [cs.AI] |
| (or arXiv:2607.02807v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.02807 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yuvraj Virk [view email]
[v1]
Thu, 2 Jul 2026 22:47:35 UTC (350 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →Procedural Memory Distillation: Online Reflection for Self-Improving Language Models
Procedural Memory Distillation (PMD) enhances reinforcement learning by converting cross-episode signals into reusable memory, improving Qwen3-8B and OLMo3-Instruct-7B models by 3.8-5.5% on SCIKNOWEVAL and 7.9-13.6% on . The co-evolution of policy and memory allows for more effective self-supervision, demonstrating significant performance gains when both components are active.