Job pack · supervisor
Multi-agent team
A second agent is a context boundary. Supervisor, crew+flows, or handoff. Never unsupervised group chat on prod tools.
When: Specialists need different tools or a fresh window, and a parent must synthesize.
Roles must change tools or schema. Fresh windows. Structured returns. Bound the talk. Prefer supervisor or handoff over group chat.
GET /api/canon/jobs/multi-agent-crew?format=md
Playbooks
- Spawn subagentsA nested loop to protect the parent window, to parallelize, or to apply a different skill set.
- Choose a topologyWalk the decision tree. Do not start from a crew because the demo looked collaborative.
- Add a human gateApproval and missing info are tools. Pause is an API. The blob survives the wait.
Doctrine to load
- Small, Focused AgentsA good agent has few tools, a narrow goal, and a short context. Generalists are orchestrators. Specialists do the work.
- Hierarchical SupervisorA supervisor node routes to worker nodes and back until a completion condition. LangGraph's most copied multi-agent topology — and the one that actually checkpoints.
- Role-Playing CrewsCrewAI's bet: give each agent a role, a goal, and a backstory, assign tasks, run a process. It is the fastest way to sketch a team — and the easiest way to overfit to theater.
- Handoffs and SwarmControl transfers. The current agent returns the next agent. Context is a clean brief, not a pile of speakers. OpenAI Swarm made this obvious; the Agents SDK made it shippable.
- Conversational Multi-AgentAutoGen's original idea: agents as speakers in a group chat, with a manager selecting the next speaker. Powerful for research; dangerous as a production bus.
- Society of MindMetaGPT's software company: SOPs as prompts, specialized roles producing artifacts for the next role. Minsky as a Makefile.
Refuse
- Shared mega-context. A crew sharing one bloated window. Specialists drown in each other's work.
- Role-play without a different ACI. Backstories that do not change tools, prompt constraints, or schema. Theater.
- Unsupervised group chat. Agents talking to each other against production tools with no turn bound and no typed messages.
- Handoff history dump. Transfer of control copies the entire conversation into the specialist.
Checklists
Ready to build
- Named termination: done schema, max turns, token/money cap.
- Sandbox / allowlist for every side effect. Secrets in the executor.
- An explicit context builder: pinned, working, recalled, compacted.
- A golden set of 20–50 real tasks, run in CI on prompt change.
- Launch / pause / resume APIs. HITL is a tool that checkpoints.
Harness quality
- Filesystem workspace. Disk is source of truth; the window is a cache.
- Progressive-disclosure skill loader.
- Plan/act modes. Plan is a file the human can edit.
- Subagent spawner with reduced tools and a written return.
- Sandbox plus permission gates.
- Verifier in the loop (test, typecheck, build, browser).
Recipes
Default corpus
langgraphcrewaiopenai-agentsdeepagents