A reading list for when one LLM isn't enough

What it does This is an “awesome list” — a hand-curated bibliography of academic papers about multi-agent LLM systems. It sorts research into three buckets: collaboration and system design, frameworks and benchmarks, and application-specific work like software engineering. Each entry is a paper title with a direct arXiv or Hugging Face link.

The interesting bit The list doubles as a map of how fast this niche is moving. Papers from 2023 (AutoGen, CAMEL) sit next to 2025 entries on “virtual agent economies” and “agentic system optimization,” suggesting the field is still defining its own vocabulary.

Key highlights

• Covers core topics: multi-agent collaboration, reasoning, tool use, and failure modes
• Includes benchmarks like AgentClinic and TheAgentCompany for evaluating real-world agent performance
• Heavy representation from arXiv and Hugging Face papers, with some Semantic Scholar and IEEE sources
• Explicitly maintained by a team building the swarms multi-agent framework, so the curation reflects their research priorities
• Mix of conceptual work (“Can AI Agents Agree?”) and systems papers (“AIOS: LLM Agent Operating System”)

Caveats

• No annotations or summaries beyond paper titles; you get links, not context
• The “Description” field promised in the format header is mostly absent
• Truncated README suggests the application sections (software engineering, etc.) continue beyond what was retrieved

Verdict Useful if you’re doing a literature review or trying to name-check the right papers in a grant proposal. Skip it if you want analysis, code, or guidance on which ideas actually work in production.