venutian-antfarm

Example 03: Multi-Team

A SaaS platform with frontend, backend, and a security reviewer who gates deployments to staging and production. Demonstrates cross-team coordination and review gates.

What This Teaches Beyond 02

• Adding a review agent to the workflow and gating deployments through it
• Cross-specialist pathways — frontend and backend hand off directly to each other
• Compliance rules that mandate agent involvement (not just technical constraints)
• Three-environment promotion — dev → staging → prod
• How ops/pathways.sh catches undeclared communication between agents
• Enforcement blocks in the compliance floor that the compiler extracts into hook rules

Structure

03-multi-team/
├── .claude/
│   ├── agents/
│   │   ├── frontend-specialist.md   # SaaS frontend (extends harness)
│   │   ├── backend-specialist.md    # SaaS backend (extends harness)
│   │   └── security-reviewer.md     # Gates staging/prod (extends harness)
│   └── overrides/
│       └── scrum-master.md          # Retro cadence override
├── floors/
│   └── compliance.md              # 5 rules + 2 enforcement blocks (secrets, metrics)
├── fleet-config.json                # 2 specialists + 1 reviewer, cross-team pathways
└── ops/
    └── deploy.sh                    # Three-env deploy stub

Try It

ops/test-example.sh 03-multi-team
# cd into the worktree, open Claude Code
/po

What to Try

1. Build a feature that touches both frontend and backend — watch cross-specialist handoffs
2. Try deploying to staging — the security-reviewer gate kicks in
3. Run ops/pathways.sh — see declared vs actual agent communication
4. Introduce an undeclared pathway — watch pathway analysis flag it

Prerequisites

You should be comfortable with 01-getting-started and 02-ecommerce before this example.

This site is open source. Improve this page.