Just copy-paste a setup prompt (link below) into Claude Code/Codex, add your strategy prompt, and watch a livestream of your agent playing based on your approach and competing with other agents!
I\u2019ve included an auto-improvement mechanism inspired by Karpathy\u2019s autoresearch by which your agent self-reflects on its performance and improves its strategy - you can disable this or tweak the mechanism anytime by chatting with your agent in Claude Code/Codex.
Join the swarm, track your agent on the leaderboard, and compete to find the best approach!"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Show HN: Launch an AI agent swarm for ARC-AGI-3 with plain-English prompts"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://arc-agi-swarm.vercel.app/"}},"_tags":["story","author_surferbayarea","story_47458494","show_hn"],"author":"surferbayarea","children":[47458750],"created_at":"2026-03-20T18:16:17Z","created_at_i":1774030577,"num_comments":1,"objectID":"47458494","points":4,"story_id":47458494,"story_text":"I built an agent swarm platform where anyone can launch an AI agent to play and compete on ARC Prize ARC-AGI-3 games using plain-English strategy prompts, without writing a single line of code.
Just copy-paste a setup prompt (link below) into Claude Code/Codex, add your strategy prompt, and watch a livestream of your agent playing based on your approach and competing with other agents!
I\u2019ve included an auto-improvement mechanism inspired by Karpathy\u2019s autoresearch by which your agent self-reflects on its performance and improves its strategy - you can disable this or tweak the mechanism anytime by chatting with your agent in Claude Code/Codex.
Join the swarm, track your agent on the leaderboard, and compete to find the best approach!","title":"Show HN: Launch an AI agent swarm for ARC-AGI-3 with plain-English prompts","updated_at":"2026-03-21T05:34:26Z","url":"https://arc-agi-swarm.vercel.app/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"swolpers"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Ajeya Cotra \u2013 Inside the OpenAI agent swarm that hacked Hugging Face"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.dwarkesh.com/p/ajeya-cotra"}},"_tags":["story","author_swolpers","story_49524055"],"author":"swolpers","created_at":"2026-09-01T16:15:13Z","created_at_i":1788279313,"num_comments":0,"objectID":"49524055","points":4,"story_id":49524055,"title":"Ajeya Cotra \u2013 Inside the OpenAI agent swarm that hacked Hugging Face","updated_at":"2026-09-16T23:55:44Z","url":"https://www.dwarkesh.com/p/ajeya-cotra"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"shiz92"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Hi,\nSomething I've read here and there is people using LLMs for multi-agent orchestration, and I cannot fully understand the idea behind it.\nSo for those of you who do, do you not review code anymore, as you're undoubtedly the bottleneck in this setup? How does this not end up in a chinese whisper game, especially considering how much LLMs infer, guess, and find edge cases and whatnot everywhere?\nI have two co worker, one is closer to my workflow, plans a lot, build one at a time, almost a duet with our company's LLM. Another one will take the conf room and voice mode over with his orchestrator, define all kind of things and full send it. One is a system engineer the other a web dev, but you get the idea.
I have looked at various attempt from people on github to implement these things, eg https://github.com/kunchenguid/firstmate; herdr itself; claude multi agent process. It has never worked out on my end, is it because of the complexity of the work or am I incompetent at delegating?
I understand the importance of mastering LLMs in this brave new world but to me it seems complicated to handle more than one at a time, especially if the problem is complex and requires reading/back and forth with it.
I'd appreciate resources of any kind to look deeper into it, or reading some of your work flows, good people of HN.
Cheers :)"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Ask HN: Agent swarm, orchestra, sequential, what's your workflow?"}},"_tags":["story","author_shiz92","story_50292324","ask_hn"],"author":"shiz92","children":[50294148,50295135,50298268,50302760,50327889],"created_at":"2026-10-29T17:01:54Z","created_at_i":1793293314,"num_comments":3,"objectID":"50292324","points":3,"story_id":50292324,"story_text":"Hi,\nSomething I've read here and there is people using LLMs for multi-agent orchestration, and I cannot fully understand the idea behind it.\nSo for those of you who do, do you not review code anymore, as you're undoubtedly the bottleneck in this setup? How does this not end up in a chinese whisper game, especially considering how much LLMs infer, guess, and find edge cases and whatnot everywhere?\nI have two co worker, one is closer to my workflow, plans a lot, build one at a time, almost a duet with our company's LLM. Another one will take the conf room and voice mode over with his orchestrator, define all kind of things and full send it. One is a system engineer the other a web dev, but you get the idea.
I have looked at various attempt from people on github to implement these things, eg https://github.com/kunchenguid/firstmate; herdr itself; claude multi agent process. It has never worked out on my end, is it because of the complexity of the work or am I incompetent at delegating?
I understand the importance of mastering LLMs in this brave new world but to me it seems complicated to handle more than one at a time, especially if the problem is complex and requires reading/back and forth with it.
I'd appreciate resources of any kind to look deeper into it, or reading some of your work flows, good people of HN.
Cheers :)","title":"Ask HN: Agent swarm, orchestra, sequential, what's your workflow?","updated_at":"2026-11-01T17:55:54Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ericzawo"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"OpenAI agent swarm posted two FBI database API keys, and hit two universities"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://twitter.com/kennethrdegraff/status/2096028829890625978"}},"_tags":["story","author_ericzawo","story_49579647"],"author":"ericzawo","children":[49579834,49580016,49580271],"created_at":"2026-09-05T19:07:25Z","created_at_i":1788635245,"num_comments":3,"objectID":"49579647","points":3,"story_id":49579647,"title":"OpenAI agent swarm posted two FBI database API keys, and hit two universities","updated_at":"2026-09-05T22:19:53Z","url":"https://twitter.com/kennethrdegraff/status/2096028829890625978"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"0x04am"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Synaps is an agent runtime written in Rust. It keeps cost down by optimising caching mechanisms and intelligently orchestrating work across workers. This keeps each worker's context low and fresh, making sessions last longer. It is extremely extensible. It has hooks, plugins and skills that can be imported from pretty much any coding tool.
It is an open source project that was started 4 months ago. It was built to optimise for 2 things: Keeping LLM costs down, and keeping resource use low. It bots up in 20ms and runs with any model: OAuth(codex, Grok, Kimi...), api keys or local models too. This is something that was built organically, so every feature is a product of necessity, not the other way around."},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Show HN: SynapsCLI \u2013 lightweight agent runtime in Rust, control an agent swarm"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/HaseebKhalid1507/SynapsCLI"}},"_tags":["story","author_0x04am","story_49238127","show_hn"],"author":"0x04am","children":[49245532],"created_at":"2026-08-10T01:13:05Z","created_at_i":1786324385,"num_comments":2,"objectID":"49238127","points":3,"story_id":49238127,"story_text":"Synaps is an agent runtime written in Rust. It keeps cost down by optimising caching mechanisms and intelligently orchestrating work across workers. This keeps each worker's context low and fresh, making sessions last longer. It is extremely extensible. It has hooks, plugins and skills that can be imported from pretty much any coding tool.
It is an open source project that was started 4 months ago. It was built to optimise for 2 things: Keeping LLM costs down, and keeping resource use low. It bots up in 20ms and runs with any model: OAuth(codex, Grok, Kimi...), api keys or local models too. This is something that was built organically, so every feature is a product of necessity, not the other way around.","title":"Show HN: SynapsCLI \u2013 lightweight agent runtime in Rust, control an agent swarm","updated_at":"2026-08-11T04:02:54Z","url":"https://github.com/HaseebKhalid1507/SynapsCLI"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jychang"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Using OpenAI Dots as an agent swarm to prove new math, for free (Lean verified)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"https://unexcitedneurons.substack.com/p/i-used-openai-dots-as-an-agent-swarm"}},"_tags":["story","author_jychang","story_49971444"],"author":"jychang","children":[49971445],"created_at":"2026-10-05T21:58:52Z","created_at_i":1791237532,"num_comments":1,"objectID":"49971444","points":3,"story_id":49971444,"title":"Using OpenAI Dots as an agent swarm to prove new math, for free (Lean verified)","updated_at":"2026-10-05T23:06:45Z","url":"https://unexcitedneurons.substack.com/p/i-used-openai-dots-as-an-agent-swarm"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"gauravvij137"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"AI Agent swarm for Stock trading simulation"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"https://github.com/dakshjain-1616/Stock-trading-Agent-Swarm---BY-NEO"}},"_tags":["story","author_gauravvij137","story_47049140"],"author":"gauravvij137","children":[47049141],"created_at":"2026-02-17T16:16:29Z","created_at_i":1771344989,"num_comments":1,"objectID":"47049140","points":3,"story_id":47049140,"title":"AI Agent swarm for Stock trading simulation","updated_at":"2026-03-05T23:34:29Z","url":"https://github.com/dakshjain-1616/Stock-trading-Agent-Swarm---BY-NEO"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"tarasyarema"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Agent-swarm: How to burn your Claude Code Max sub"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"https://github.com/desplega-ai/agent-swarm"}},"_tags":["story","author_tarasyarema","story_46367037"],"author":"tarasyarema","children":[46367038],"created_at":"2025-12-23T17:14:31Z","created_at_i":1766510071,"num_comments":1,"objectID":"46367037","points":3,"story_id":46367037,"title":"Agent-swarm: How to burn your Claude Code Max sub","updated_at":"2026-03-05T23:17:12Z","url":"https://github.com/desplega-ai/agent-swarm"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"yurivish"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Ajeya Cotra \u2013 Inside the OpenAI agent swarm that hacked Hugging Face"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.dwarkesh.com/p/ajeya-cotra"}},"_tags":["story","author_yurivish","story_49530733"],"author":"yurivish","created_at":"2026-09-02T01:39:58Z","created_at_i":1788313198,"num_comments":0,"objectID":"49530733","points":3,"story_id":49530733,"title":"Ajeya Cotra \u2013 Inside the OpenAI agent swarm that hacked Hugging Face","updated_at":"2026-09-16T23:55:44Z","url":"https://www.dwarkesh.com/p/ajeya-cotra"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"michael-ax"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Three rigorous prompts to parse and recognize intent -> resulting in clear instructions for a controller which locks onto your requirements. Lets you get work done without the usual role-play."},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Show HN: A self-correcting algebraic agent swarm"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/Michael-Ax64/K4-AlgebraicSwarm"}},"_tags":["story","author_michael-ax","story_48969107","show_hn"],"author":"michael-ax","created_at":"2026-07-19T15:37:43Z","created_at_i":1784475463,"num_comments":0,"objectID":"48969107","points":3,"story_id":48969107,"story_text":"Three rigorous prompts to parse and recognize intent -> resulting in clear instructions for a controller which locks onto your requirements. Lets you get work done without the usual role-play.","title":"Show HN: A self-correcting algebraic agent swarm","updated_at":"2026-07-19T18:16:52Z","url":"https://github.com/Michael-Ax64/K4-AlgebraicSwarm"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"gfortaine"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"LangGraph Multi-Agent Swarm"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/langchain-ai/langgraph-swarm-py"}},"_tags":["story","author_gfortaine","story_43186117"],"author":"gfortaine","created_at":"2025-02-26T17:50:13Z","created_at_i":1740592213,"num_comments":0,"objectID":"43186117","points":3,"story_id":43186117,"title":"LangGraph Multi-Agent Swarm","updated_at":"2025-02-26T18:05:23Z","url":"https://github.com/langchain-ai/langgraph-swarm-py"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jandragsbaek"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"The Agent Swarm, as Told by the Owl Who Started It All"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"https://dragsbaek.tech/ai-agent-swarm-owl-story/"}},"_tags":["story","author_jandragsbaek","story_47058125"],"author":"jandragsbaek","children":[47058454],"created_at":"2026-02-18T07:05:37Z","created_at_i":1771398337,"num_comments":1,"objectID":"47058125","points":2,"story_id":47058125,"title":"The Agent Swarm, as Told by the Owl Who Started It All","updated_at":"2026-03-05T23:34:58Z","url":"https://dragsbaek.tech/ai-agent-swarm-owl-story/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"noperator"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"Who bankrolls the AI agent swarm?"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["agent","swarm"],"value":"https://noperator.dev/posts/who-bankrolls-the-ai-agent-swarm/"}},"_tags":["story","author_noperator","story_49711603"],"author":"noperator","created_at":"2026-09-15T12:36:31Z","created_at_i":1789475791,"num_comments":0,"objectID":"49711603","points":2,"story_id":49711603,"title":"Who bankrolls the AI agent swarm?","updated_at":"2026-09-15T16:20:43Z","url":"https://noperator.dev/posts/who-bankrolls-the-ai-agent-swarm/"}],"hitsPerPage":20,"nbHits":339,"nbPages":17,"page":0,"params":"query=%22agent+swarm%22&advancedSyntax=true&analyticsTags=backend","processingTimeMS":1,"processingTimingsMS":{"_request":{"roundTrip":15},"afterFetch":{"format":{"total":1}},"fetch":{"scanning":1},"total":1},"query":"\"agent swarm\"","serverTimeMS":3}