Skip to content

4nrry/agent-fleet

v1.5.0MIT

Rules for multi-agent work: model routing per phase, fan-out sizing, closed output schemas and citation checks, with CLI workers for search and extract and hooks that load the rules and check routing at dispatch.

GitHubOpen Website
agent-fleet

Rules for multi-agent work — routing each phase to the right model, sizing a fan-out, constraining output schemas so budget is not spent on text you discard, and verifying what agents return. Read this before any deep-research run, whether invoked by name or triggered implicitly by a question that needs many sources, and before launching subagents, a dynamic workflow, an agent team, or any parallel run. Read it again when the results come back and you must decide what to trust. Consult it even when the user never says "deep research", "fleet", "fan-out" or "subagent" — any time work is about to be split across several agents, or a batch of agent findings needs checking.