Library/@admin/agent-swarm

Agent Swarm

Multi-agent swarm — a supervisor decomposes your task into sub-tasks, runs them in parallel with shell / file / web tools, then reconciles into a single answer. Model-agnostic; swap in Kimi K2.6 the day Gonka exposes it.

jobagentv0.1.7Featured
Updated 21d ago · 32 total runs

Inputs

taskstringrequiredThe high-level goal for the swarm (one paragraph — the supervisor decomposes from here).
modelstringUpstream model id passed verbatim to proxy.gonka.gg. Today the full HF id for Qwen3 235B. When Kimi K2.5 / K2.6 land, swap in moonshotai/kimi-k2.5 etc.
max_subagentsintegerMax parallel workers per round (1-30). Higher = broader fan-out, more tokens.
budget_tokensintegerSoft token budget across the whole swarm (informational; actual hard-cap is the run-level spend cap).

Outputs

resultstring
planjson
subagent_runsjson
tokens_usedinteger

Resources

CPU
1
Memory
2048 MB
Timeout
1800s