🚀 GOAP: Agentic QE Fleet Enhancement - AG-UI & Protocol Modernization #177
Description
Executive Summary
Strategic enhancement plan for the Agentic QE Fleet to support standardized agent protocols (AG-UI, A2A, A2UI) while maintaining backward compatibility and enhancing scalability to 100+ concurrent agents.
Reference: docs/plans/goap-fleet-enhancement-2025.md
Current State (v2.7.0)
| Component | Count |
|---|---|
| Core QE Agents | 21 |
| n8n Workflow Agents | 15 |
| Subagents | 11 |
| MCP Tools | 91+ |
| Context Reduction | 87% |
Identified Gaps:
- No AG-UI integration (custom WebSocket only)
- Limited streaming semantics (non-standard events)
- Custom gossip protocol (not A2A aligned)
- No declarative UI capability (A2UI pattern)
- 50+ agent scalability ceiling
Milestone Roadmap
Milestone 1: Communication Foundation (v2.8.0)
-
Action 1.1: AG-UI Protocol Adapter (complexity: 3)
- AG-UI event wrapper for existing streaming types
- SSE endpoint alongside WebSocket
- Event taxonomy aligned with AG-UI spec
-
Action 1.2: Bidirectional Streaming (complexity: 3)
- Client interrupts/cancellation support
- State synchronization messages
- Backpressure handling
-
Action 4.2: Distributed Tracing (complexity: 3)
- Trace propagation across all 21 agents
- Jaeger/Zipkin export
- Correlation with test results
Total Complexity: 9 points | Risk: Low-Medium
Milestone 2: Scale & Visibility (v2.9.0)
-
Action 3.1: 100+ Agent Coordination (complexity: 6)
- Hierarchical coordination at 100 agents
- Memory footprint under 4GB
- Coordination latency under 100ms
-
Action 4.1: Enhanced Visualization (complexity: 3)
- Agent dependency graph with real-time updates
- Test execution heatmaps
- Memory usage sparklines
Total Complexity: 9 points | Risk: Medium-High
Milestone 3: Protocol Alignment (v2.10.0)
-
Action 2.1: A2A Protocol Alignment (complexity: 5)
- Agent capability cards (A2A format)
- Task negotiation protocol
- Agent discovery mechanism
-
Action 2.2: Message Standardization (complexity: 3)
- JSON-RPC 2.0 message envelope
- Schema validation with Ajv
- Message versioning
-
Action 4.3: Simplified Agent Spawning (complexity: 2)
aqe spawn <agent-type>CLI command- Agent templates with defaults
- Auto-coordination enrollment
Total Complexity: 10 points | Risk: Medium
Milestone 4: Advanced Patterns (v3.0.0)
-
Action 1.3: Event-Driven Architecture (complexity: 4)
- Central event bus with pub/sub
- Event replay capability
- Event persistence for debugging
-
Action 3.2: Resource Allocation Intelligence (complexity: 4)
- ML-based task complexity estimation
- Dynamic agent spawning
- Memory quota enforcement
-
Action 3.3: Distributed Memory Patterns (complexity: 5)
- CRDT-based shared state
- Memory eviction policies
- Cross-partition memory access
-
Action 2.3: A2UI Declarative UI (complexity: 5)
- Component catalog (cards, tables, charts)
- Native rendering in dashboard
- Accessibility compliance
Total Complexity: 18 points | Risk: Medium-High
Success Metrics
| Metric | Current | Target |
|---|---|---|
| Agent scalability | 50 | 100+ |
| Streaming latency | 500ms | 100ms p95 |
| Context reduction | 87% | 90% |
| Test coverage | 85%+ | 90%+ |
Risk Matrix
| Risk | Probability | Mitigation |
|---|---|---|
| Protocol Volatility (AG-UI/A2A evolving) | Medium | Adapter layers, not rewrites |
| Backward Compatibility | Medium | Feature flags, gradual rollout |
| Performance Regression | Low | Benchmark gates in CI |
| Resource Constraints | High | Continue batched testing policy |
References
Total Path Cost: 46 complexity points
Estimated Milestones: 4 release cycles (v2.8.0 → v3.0.0)
🤖 Generated with Claude Code