skills/proffesor-for-testing__agentic-qe

mirror of https://github.com/proffesor-for-testing/agentic-qe.git synced 2026-09-19 08:45:47 +08:00

Files

T

Profa 7abb1698e1 chore(release): bump version to v3.6.1

ADR-064 Agent Teams integration, distributed tracing, competing hypotheses,
dynamic scaling, federation mailbox, circuit breakers, task DAG scheduling,
HNSW graph indexing, pattern training pipeline, and Devils Advocate agent.

Also restructured README with separate release notes, fixed release skill
with real CLI commands and npm scripts.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-09 19:53:36 +00:00

2.5 KiB

Raw Permalink Blame History

v3.6.1 Release Notes

Release Date: 2026-02-09

Highlights

Agent Teams Integration (ADR-064) delivers a hybrid fleet architecture layering Claude Code Agent Teams communication on the Queen Coordinator.

Added

Agent Teams Integration (ADR-064) — Hybrid fleet architecture with direct mailbox messaging, domain-scoped teams, and tiered fleet activation (Lite/Standard/Full)
Distributed Tracing — TraceContext propagation across agent messages via correlationId encoding. Queen traces task lifecycles end-to-end with span start/complete/fail
Competing Hypotheses — Multi-agent root cause investigation with evidence scoring and convergence. Auto-triggered on critical (P0/P1) task failures
Dynamic Agent Scaling — Workload-based auto-scaling integrated into Queen's metrics loop with configurable policies and cooldowns
Cross-Fleet Federation — FederationMailbox for multi-service routing with health monitoring and service discovery
Domain Circuit Breakers — Fault isolation per domain with configurable thresholds and automatic recovery
Task Dependency DAG — Topological ordering with cycle detection for complex multi-step workflows
HNSW Graph Construction — Real O(log n) HNSW insert/search replacing the O(n) stub in unified memory
Pattern Training Pipeline — TaskCompleted hook automatically extracts and trains patterns into ReasoningBank
Devils Advocate Agent — qe-devils-advocate for adversarial review of agent outputs

Fixed

6 CodeQL alerts resolved in enterprise-integration services
TraceContext now correctly propagated through adapter sendMessage/broadcast
Dynamic scaler wired into Queen's metrics collection loop
Competing hypotheses auto-triggered on permanent P0/P1 failures

Changed

Queen Coordinator now manages trace contexts for all task lifecycles
Agent Teams adapter supports trace-aware messaging
Fleet tier descriptions updated for Lite/Standard/Full activation levels

Architecture

ADR-064: Agent Teams Integration — 4-phase hybrid architecture
- Phase 1: Foundation (AgentTeamsAdapter, FederationMailbox)
- Phase 2: Hybrid Architecture (CircuitBreaker, TaskDAG, DynamicScaler)
- Phase 3: Learning Pipeline (ReasoningBankPatternStore, promotePattern)
- Phase 4: Observability (DistributedTracing, CompetingHypotheses)

Test Coverage

282 coordination tests passing
397 total tests across 18 test files
8 new tracing benchmark tests (all <500ms, typically <10ms)

2.5 KiB Raw Permalink Blame History