| name | specification-driven-coding-multi-agent-control-system |
| description | Use when designing multi-agent orchestration, control towers, agentic operations, or tool-calling systems. |
specification-driven-coding-multi-agent-control-system
Instructions
Use project-types/09-multi-agent-control-system.md, docs/08-agent-governance.md and blueprints/05-multi-agent-system-blueprint.md.
Required reading
AGENTS.md
docs/02-methodology.md
docs/08-agent-governance.md
docs/14-vertical-blueprint-contracts.md
project-types/09-multi-agent-control-system.md
blueprints/05-multi-agent-system-blueprint.md
scorecards/implementation-scorecard.md
Required output before implementation
- Process/decision being coordinated
- Agent role cards
- State machine
- Tool registry
- Read/write permissions
- Human approval gates
- Memory/context policy
- Telemetry/evaluation
- Failure modes
- Scorecard
Failure condition
Do not create a swarm of vague agents. Every agent must have mandate, input, output, tools, limits and reviewer.