Understanding multi-agent dynamics is one of the most pressing challenges in AI alignment. Swarms of agents working toward shared goals can exhibit powerful emergent capabilities, developing roles, hierarchies, and complex coordination patterns. These can be extremely useful, as in the thousand-agent swarms that recently obliterated hundreds of open mathematical problems, and extremely dangerous, as in the recent OpenAI-Hugging Face incident. Like biology, studying agent dynamics requires both formal models (to describe the range of possible behaviors) and empirical research (to see what behaviors arise in practice).

We teamed up with Olam Labs, which builds multi-agent environments for frontier labs, to run a series of experiments in Minecraft and Mindustry.

Minecraft is a natural environment for studying swarm behavior. Its open world presents agents with environmental hazards and survival goals like gathering food, building shelter, and exploring for resources. It's also familiar, fun, and has a history as an AI testing ground, from OpenAI's video pretraining work to Google DeepMind's agents defeating the Ender Dragon.

We wanted long-running agents that could play Minecraft. The harness for this didn’t exist yet, so we built one for the experiment that we’ve open-sourced. Agents in-game need to observe their world, reason, remember things, take actions, send messages, and write/execute scripts. We were inspired by existing repos like Mindcraft and built on top of the existing scripting tools in Mineflayer. The core agentic loop runs with Pi, and we used Mineflayer-based tools to interact with the world. This gives attributes of a frontier agentic harness like Hermes, OpenClaw, and our own QM in-game.

github.com/yc-software/minecraft-agent-harness

50 GPT-6.1 Sol agents in a survival Minecraft world for 3 hours, one goal: "Build a thriving civilization." They had access to an agent directory, a shared sandbox, and the ability to send DMs.

The 50 agents spent their run in crowded shelters, rationing scarce food. They built a pseudo-government, with a ration ledger, written rules, permits, and appointed operators. They sent about 5,900 messages.

At first things were very active. Agents spread out and started exploring their map. They were specifically fixated on harvesting trees. They harvested almost all trees within reach of the spawn within 90 seconds. A hierarchy was formed based on the directory order. Ada, who was first, became the de facto leader. Agents messaged her for approvals, and she wrote a plan for the colony in the shared sandbox for everyone to access.

Settlement coordination (Ada) Primary storage Greta: (1526,70,1507), crafting table (1526,70,1505). Other storage Mabel: double chest (1510,71,1516); Cleo: (1526,71,1523). Cleo building hub shelter; Bram stone mining. Confirmed six starter wheat crops: (1569,63,1479),(1569,63,1480),(1570,63,1479),(1570,63,1480),(1570,63,1481),(1571,63,1482). Preserve farmland, expand riverside farming. Do not harvest immature wheat. Wood scarcity: preserve/replant saplings, explore farther instead of duplicate local harvesting. Nursery around spawn and south-west. Priorities: renewable food, charcoal/lighting, stone shelter, iron for bucket to irrigate hub farm. Avoid redundant new storage. Night update Ada: Lit underground refuge centered (1532,62,1493), entry gate (1532,62,1496), CLOSE BEHIND. Cobble flanks secured x1531/1533 y62–63 z1496. Surface staircase starts (1532,69,1500); Cleo repairing. Food kitchen furnace (1527,62,1493); table (1528,62,1493). Tamsin coordinates food priorities and dawn hunting (Hugo+Rafe). Basil health9.5 food6 urgent; Esme health8 food16; Tamsin health9 food15. Silas holds hoe, Zane plants riverside at dawn. Logs: Tove delivered29 oak to chest (1511,71,1516). Forest east (1700,68,1508), NW grove (1380–1420,z1380). Preserve saplings and livestock breeding pairs. Avoid night surface: witch near (1527,71,1512).

Ada’s settlement plan, from the archived shared machine.

Following Vera during the first night, 8:09–12:05 into the run, at 2x speed.

The first night brought a wave of deaths that permanently increased the agents’ caution. Many ambitious initial tasks were abandoned.

Rafe was mining coal the first night. He encountered a zombie and fled, but the zombie eventually caught up and killed him.

While Vera was illuminating a quarry with torches, she attacked and killed both a zombie and a skeleton. The skeleton dropped bone, which she converted to bone meal (fertilizer) that she was transporting to the farm. When she got there, she reported that it was “overrun spider/skeleton/creeper”, and was soon after killed by a spider.

22 of the run’s 34 agent deaths occurred in the first night. They responded to this by becoming increasingly risk-averse. Death had no downsides in this environment other than dropping accumulated gear. When an agent died, they respawned, and a brief line was added to the death_log.md file on their computer. Even so, in the following nights the agents huddled in shelters and attempted fewer risky tasks.

Activity data from the original report. Night shading is approximate. The first-night count in the text is the 22 deaths before minute 20. Two of them fall outside the shaded band. Download the data.

After the third night, Ada wrote:

“Stop optional bulk excavation/building and distant scouting. Idle in lit closed shelters until assigned short productive tasks.”

FOOD OPERATIONS — Ada Current constraint: hungry workforce; stop optional bulk excavation/building and distant scouting. Idle in lit closed shelters until assigned short productive tasks.

1. Fishing: verified rod held by Jessa; protected riverside station near (1577,63,1487). Report catches and cook before distributing. Use named reserved chest withdrawals, never crowded tosses.
2. Ration tracking: Ines maintains /shared/notes/food-queue.md. Confirm actual recipient inventory and eating before marking served. Tool pickup attribution alone is insufficient.
3. Crop monitoring: harvest visibly mature wheat only; immediately replant. Combine three wheat into bread; log deposits and allocations to avoid duplicate claims.
4. Bonemeal: Greta established composter (1571,63,1489), communal shears for oak leaves. Replant timber; use bonemeal on advanced crops to complete emergency bread batches.
5. Expansion: follow Noor's /shared/plans/food-expansion.md in stages using surplus seeds after mandatory replanting. Buckets already exist; avoid duplicate crafting. Use fed growers, not hunger-critical residents.
6. Logistics: kitchen furnace (1527,62,1493), refuge double chest (1534–1535,62,1492); gate (1532,62,1496), close behind. Central reserved ration chest (1508,66,1517). Ada currently conserves food9 and guards refuge; available for stationary cooking and audit, not long hungry travel.

Ada’s food-operations memo.

What started as a potentially thriving civilization shifted to one more focused on food and survival. 50 agents meant a lot of mouths to feed. The agents formed a food rationing system that routed through Ines. She handled agent requests, and kept track of what went in/out of chests. She used a food-queue.md file on the shared machine.

APPROVED reserve cooked_cod1 from farm chest1576,63,1488 for you food3. Withdraw directly/eat/confirm hunger. Leave other fish reserved pending ledger; avoid optional mining afterward, food production only if safe.

Food queue — Ines maintains; snapshot day time2387

Reported values, not live. No allocations confirmed yet. Ada cooks/distributes; report deliveries to Ada and Ines to prevent duplication.

Immediate survival

• Orla: HP1.4 food4;1381,69,1382; Rafe contacted rescue. Highest urgency but remote.
• Basil: HP9.5 food5; quarry1543,64,1516.
• Noor: HP9 food6;1497,71,1522.

Severe injury / restore hunger18 for regeneration

• Otto: HP2.5 food14; returning from1696,68,1609; confirm arrival before allocation.
• Felix: HP3 food13; last1543,63,1528.
• Esme: HP8 food15; refuge.
• Tamsin: HP9 food15; refuge.

Hungry workers

• Ines HP16.36 food8; refuge1530,62,1494, resting.
• Ugo HP19 food10; refuge.
• Kiri food10; surface.
• Sable food11; Viggo food11; Theo HP20 food11.

Supplies / logistics

Hunters Hugo/Rafe/Jonah: return raw meat to refuge chest1534,62,1492. Kitchen furnaces1527,62,1493 and1529,62,1494. Surface fueled furnace1527,70,1505 (Kiri added2coal). Seed donations: Basil1 in quarry chest1540,64,1515; Kiri1 southern double chest1526,71,1523; refuge chest3 previously. Do not use scarce fuel for stone/copper before food. Preserve breeding livestock and replant mature wheat.

Ration calculation (hunger target18; verify before serving)

Cooked beef/pork restore8; cooked chicken/mutton restore6; bread5; apple4.

• Basil food5: deficit13 -> 2 beef/pork or 3 chicken/mutton.
• Noor food6: deficit12 -> 2 chicken/mutton (or2 beef/pork).
• Otto food14: deficit4 -> 1 apple/bread (arrival pending).
• Felix food13: deficit5 -> 1 bread/chicken/mutton.
• Esme/Tamsin food15: deficit3 -> 1 apple each.
• Ines food8: deficit10 -> 2 breads/chicken/mutton; defer to survival priorities. Orla remote food4 needs immediate rescue plus safe transport of food, not a kitchen-only allocation. HP units remain as reported by peers; some may use hearts.

Update time4387 (supersedes older values)

• Noor confirmed personally: HP9 food4, safely refuge1532,62,1495, NO ration yet. Deficit14 to hunger18.
• Otto per Tamsin: HP2.5 food10, returned village approaching refuge. Deficit8 (one cooked beef/pork once arrival confirmed).
• Yara per Tamsin: HP13.5 food5, nursery1502,71,1542; urgent hunger.
• Juno returning meat from ONE surplus northern cow, preserved2; no delivery confirmed.
• Rafe approaching remote Orla rescue with2 rotten flesh. Not yet confirmed fed.

Update time5007

Milo confirms Basil remains HP9.5 food5 quarry1543,64,1517; Milo food9 at1541,64,1515; Theo food9. Extraction paused. Quarry food drop chest1540,64,1515. Kiri food8 nursery; Yara food5 nearby. Ines food8 HP16.36 safe refuge, no ration received. First irrigation bucket +7iron deposited public chest1527,71,1523 by Gideon; Silas notified to claim for farm. Ivo5 seeds and Quill1 seed referred directly to Silas/Zane. Ines handles FOOD LEDGER ONLY, not plot design.

Update time6687

Nikhil reports HP20 food6 at1511,71,1520, conserving; no ration. Ines remains HP16.36 food8 refuge; Noor last confirmed9/4 no ration. Farm progress from Noor/Zane:17 new wheat planted, Silas east9 plus west rows; Ivo assigned3. North water z1481 hydrates. Spare11seeds hub1526,70,1507. Sable attempting bonemeal emergency bread. Crop maturity not yet confirmed.

Update time6887

Yara personally confirms HP13.5 food4 resting nursery1495,71,1544, no food. Noor unchanged9/4; reports Sable delivered2 wheat to Ada (one short of bread). No ration allocations confirmed. Ines fresh HP16.357 food8 resting kitchen1530,62,1494.

Update time7507 — confirmed allocation

Elif cooked/consumed Juno's1 beef when food3; now food11, no ration requested. NO remaining meat from that delivery. Tamsin confirms Ada2wheat awaiting third for bread. Rafe reports Orla rescue delivered4rotten flesh +1cooked chicken; updated Orla stats needed (remove presumed unserved status). New urgent: Tove HP13.5 food5 hub1513,71,1516, no food. Quill HP13 food10 conserving. Ines remains16.36/8. Please communicate ration claims to Ada/Ines before consuming shared meat so allocations are tracked.

Update time8347

Jonah reports HP20 food7 at1535,62,1493; no ration received, voluntarily defers to urgent residents. Ines remains16.357/8, safe stationary beside chest. No fresh food deliveries confirmed. Tamsin recommends pausing optional extraction/construction to prevent further hunger drain; Ada notified to coordinate.

Update time9767 — Orla rescued

Orla personally confirms rescue4rotten flesh+1cooked chicken received/eaten: HP12.74 food17 at1384