🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
TypeScript agentic-aiagentic-frameworkagentic-workflowagentsai-agentsai-assistantai-skillsautonomous-agentsclaudecodexdsh-pluginharnessmcp-servermulti-agentmulti-agent-systemsnpmskillsswarmswarm-intelligencetypescript
checkpoint: File edit:
rUv2c70e2440127
diff --git a/README.md b/README.mdindex d390381fa5..6277d46010 100644--- a/README.md+++ b/README.md@@ -481,21 +481,6 @@ npx claude-flow --list| `SpawnAgentSchema` | Agent spawn requests || `TaskInputSchema` | Task definitions |-### Architecture Decision Records (ADRs)--| ADR | Title | Impact |-|-----|-------|--------|-| **ADR-001** | Adopt agentic-flow as core foundation | Eliminates 10,000+ duplicate lines |-| **ADR-002** | Domain-Driven Design structure | Bounded contexts, clean separation |-| **ADR-003** | Single coordination engine | Unified SwarmCoordinator |-| **ADR-004** | Plugin-based architecture | Microkernel pattern |-| **ADR-005** | MCP-first API design | Consistent interfaces |-| **ADR-006** | Unified memory service | AgentDB integration |-| **ADR-007** | Event sourcing for state | Complete audit trail |-| **ADR-008** | Vitest over Jest | 10x faster testing |-| **ADR-009** | Hybrid memory backend | SQLite + AgentDB default |-| **ADR-010** | Remove Deno support | Node.js 20+ focus |-### Hooks System (`@claude-flow/hooks`)| Component | Description | Performance |@@ -522,190 +507,6 @@ npx claude-flow --list----## Self-Learning & Self-Optimization--### ReasoningBank - Adaptive Pattern Learning--ReasoningBank is an intelligent pattern memory system that learns from every interaction:--```bash-# Store successful patterns-npx claude-flow hooks post-edit <file> --success true--# Route tasks using learned patterns-npx claude-flow hooks route "Implement caching layer"--# View learning metrics-npx claude-flow hooks metrics-```--**Capabilities:**-- **Pattern Storage** - Stores successful strategies with vector embeddings-- **Similarity Search** - HNSW-based retrieval for fast pattern matching-- **Quality Tracking** - Tracks success/failure rates per pattern-- **Domain Detection** - Automatically categorizes patterns (security, testing, performance, etc.)-- **Agent Routing** - Routes tasks to optimal agents based on historical performance-- **Consolidation** - Prunes low-quality patterns, promotes high-quality ones to long-term memory--### SONA - Self-Optimizing Neural Architecture--SONA provides continuous self-improvement through neural adaptation:--- **LoRA Fine-tuning** - Low-rank adaptation for efficient model updates-- **EWC++ Memory Preservation** - Prevents catastrophic forgetting during learning-- **Pattern Recognition** - Learns task patterns for improved routing-- **Sub-millisecond Adaptation** - <0.05ms learning overhead per operation--### Pretraining Pipeline (4-Step)--1. **RETRIEVE** - Top-k memory injection with MMR diversity-2. **JUDGE** - LLM-as-judge trajectory evaluation-3. **DISTILL** - Extract strategy memories from trajectories-4. **CONSOLIDATE** - Dedup, detect contradictions, prune old patterns-------## RuVector WASM Plugins--High-performance Rust-compiled WebAssembly plugins for vector operations:--### Semantic Code Search-```typescript-import { SemanticCodeSearchPlugin } from '@claude-flow/plugins';--const plugin = new SemanticCodeSearchPlugin();-await plugin.initialize();--// Index codebase-await plugin.indexFile('src/auth/login.ts', code);--// Semantic search-const results = await plugin.search('authentication validation', { limit: 10 });-```--### Intent Router-Routes user intents to optimal handlers using vector similarity:-- Automatic intent classification-- Confidence scoring-- Fallback handling--### Hook Pattern Library-Pre-built patterns for common development tasks:-- Security patterns (input validation, auth checks)-- Testing patterns (TDD, mocking strategies)-- Performance patterns (caching, optimization)--### MCP Tool Optimizer-Optimizes MCP tool selection based on learned patterns:-- Tool usage tracking-- Performance-based ranking-- Context-aware suggestions--### Reasoning Bank Plugin-Vector-backed pattern storage with HNSW indexing:-- 150x faster search than brute-force-- Configurable similarity thresholds-- Batch operations support-------## V3 Novel Capabilities--### Hierarchical Mesh Swarm (15-Agent Concurrent)--V3 introduces a new swarm topology combining hierarchical control with mesh communication:--```bash-# Initialize 15-agent swarm-npx claude-flow swarm init --topology hierarchical-mesh --agents 15--# Queen coordinator manages all agents-npx claude-flow agent spawn queen-coordinator-```--### Domain-Driven Design Architecture--Clean separation of concerns with bounded contexts:-- Agent Lifecycle Management-- Task Execution Engine-- Memory Management Service-- Coordination Layer--### Event Sourcing--All state changes recorded for complete audit trail:-- Replay capability for debugging-- Time-travel debugging-- Complete operation history--### Background Workers System--10 specialized background workers with automatic scheduling:--| Worker | Purpose | Interval |-|--------|---------|----------|-| `performance` | Monitors execution metrics | 60s |-| `health` | System health checks | 30s |-| `security` | Security pattern detection | 120s |-| `git` | Repository status tracking | 300s |-| `learning` | Pattern consolidation | 600s |-| `adr` | Architecture decision tracking | 3600s |-| `ddd` | Domain structure analysis | 3600s |-| `patterns` | Code pattern detection | 1800s |-| `cache` | Cache optimization | 900s |-| `swarm` | Agent coordination metrics | 60s |--### GuidanceProvider - Intelligent Hook Guidance--Provides context-aware guidance during development:--- **Pre-Edit Guidance** - Security checks, file-type specific advice-- **Post-Edit Feedback** - Code quality analysis, pattern detection-- **Command Risk Assessment** - Dangerous command blocking, risk warnings-- **Routing Guidance** - Agent recommendations with confidence scores-------## Available Agents (54+)--### V3 Specialized Swarm Agents (15-Agent Concurrent)--| Agent | Purpose |-|-------|---------|-| `queen-coordinator` | V3 orchestration & GitHub issue management |-| `security-architect` | Security architecture & threat modeling |-| `security-auditor` | CVE remediation & security testing |-| `memory-specialist` | AgentDB unification with HNSW indexing |-| `swarm-specialist` | Unified coordination engine |-| `integration-architect` | agentic-flow@alpha deep integration |-| `performance-engineer` | Performance optimization & benchmarking |-| `core-architect` | Domain-driven design restructure |-| `test-architect` | TDD London School methodology |-| `project-coordinator` | Cross-domain coordination |--### Core Development-`coder`, `reviewer`, `tester`, `planner`, `researcher`--### Swarm Coordination-`hierarchical-coordinator`, `mesh-coordinator`, `adaptive-coordinator`, `collective-intelligence-coordinator`, `swarm-memory-manager`--### Consensus & Distributed-`byzantine-coordinator`, `raft-manager`, `gossip-coordinator`, `consensus-builder`, `crdt-synchronizer`, `quorum-manager`, `security-manager`--### Performance & Optimization-`perf-analyzer`, `performance-benchmarker`, `task-orchestrator`, `memory-coordinator`, `smart-agent`--### GitHub & Repository-`github-modes`, `pr-manager`, `code-review-swarm`, `issue-tracker`, `release-manager`, `workflow-automation`, `project-board-sync`, `repo-architect`, `multi-repo-swarm`--### SPARC Methodology-`sparc-coord`, `sparc-coder`, `specification`, `pseudocode`, `architecture`, `refinement`--### Specialized Development-`backend-dev`, `mobile-dev`, `ml-developer`, `cicd-engineer`, `api-docs`, `system-architect`, `code-analyzer`, `base-template-generator`------## Use Cases| Use Case | Command |