Features

Key Features

• ReAct Engine — DefaultReActAgent implements the full Reasoning-Acting cycle with configurable iteration limits, streaming responses, and multi-layer error recovery
• 6-Stage Context Compaction — Progressive pipeline (Snip → Micro → Collapse → Auto → Partial → CircuitBreaker) with "Facts First" strategy to preserve raw context as long as possible
• 21 Built-in Tools — File ops (Read/Write/Edit/Glob/Grep), execution (Bash/Monitor), interaction (AskUser), skills (SkillList/SkillLoad), and agent ops (Spawn/Message/Task/Team/Plan)
• Read/Write Partition — READ_ONLY tools execute in parallel, WRITE/SYSTEM_CHANGE tools serialize automatically
• Human-in-the-Loop — Three-state permission model (ALLOWED/ASK/DENIED) with PermissionGuard
• Multi-Agent Orchestration — TaskBoard, PlanBuilder, TeamScheduler, and in-process MessageBus
• A2A Protocol — Agent-to-Agent communication standard (Google ADK-compatible), in-process discovery + invocation, team auto-registration
• Middleware Pipeline — Declarative request/response interception with @MiddlewareOrder for cross-cutting concerns (logging, auth, rate-limiting)
• Agent Snapshot/Checkpoint — Serialize agent state mid-conversation, restore from checkpoint with AgentBuilder.restoreFrom(snapshot)
• Structured Output — Call models returning typed POJOs with automatic self-correction on format errors
• Hook Lifecycle — 10 hook points (Pre/Post Reasoning, Acting, etc.) with CONTINUE/MODIFY/SKIP/ABORT/INJECT decisions
• Circuit Breaker — Three-state circuit breaker for both model and tool calls with configurable thresholds
• Loop Detection — Hash-based + frequency-based dual detection to prevent infinite agent loops
• Cooperative Cancellation — Graceful agent termination with state preservation
• MCP Integration — StreamableHTTP + Elicitation Protocol for external tool server connectivity
• Skill System — Markdown-based skill definitions with TriggerGuard anti-contamination design
• Plan Mode — Separate planning from execution; write tools blocked during planning
• Model Harness — Deep Anthropic integration + OpenAI-compatible fallback for GLM, Qwen, GPT, etc.
• Session Persistence — File-based state serialization with TTL cleanup

Model Support

Provider	Models	API Type	Environment Variable
Anthropic	Claude Sonnet, Claude Opus, Claude Haiku	Native Anthropic API	ANTHROPIC_API_KEY
Zhipu AI	GLM-4-Plus, GLM-4	OpenAI-compatible	GLM_API_KEY
DashScope	Qwen-Plus, Qwen-Max, Qwen-Turbo	OpenAI-compatible	QWEN_API_KEY
OpenAI	GPT-4o, GPT-4, GPT-3.5	OpenAI-compatible	OPENAI_API_KEY

// Anthropic (native API)
AnthropicProvider claude = new AnthropicProvider(apiKey);

// GLM / Qwen / GPT (OpenAI-compatible)
OpenAIProvider provider = new OpenAIProvider(apiKey, baseUrl, "/chat/completions");

Demo Examples

Demo	API Key	What it tests
AgentExample --mock	No	Basic ReAct loop with mock model
AgentExample --glm	GLM	ReAct loop with GLM-4-Plus
AgentExample --qwen	Qwen	ReAct loop with Qwen-Plus
FullToolsetExample	Qwen	All 6 tools: read, write, edit, glob, grep, bash
SkillExample	Qwen	Skill system: list, load, and use Markdown skills
MultiAgentExample	No	TaskBoard DAG tracking + MessageBus pub/sub
SessionExample	No	FileMemoryStore + SessionSerializer round-trip
Spring Boot Demo	Yes	REST API, streaming, structured output, hooks, MCP