How openjiuwen-ai/agent-core works
openJiuwen Core is a Python SDK and runtime for building LLM agents. A developer configures an agent and workflow; a user request enters the runtime, where an agent controller plans or advances workflow tasks, invokes model and tool capabilities, and returns a result. Sessions and context support continuity and interruption recovery. The README also highlights multi-agent/workflow orchestration, prompt optimization, and observability. Some links below reflect README-level capabilities rather than inspected call sites; source excerpts are partial, so uncertain internal wiring is omitted.
Main entry points: llm_controller.py ReAct Controller · runner Agent Runtime · core Workflow Orchestration · application Agent Interfaces
Components 17
Agent SDK
- Agent Interfaces
openjiuwen/core/application
Execution Runtime
- ReAct Controller
openjiuwen/core/application/llm_agent/llm_controller.py - Workflow Orchestration
openjiuwen/core - Agent Runtime
openjiuwen/core/runner - LLM Invocation
openjiuwen/core/foundation/llm - Tool Calling
openjiuwen/core - Streaming Execution
openjiuwen/core - Multi-Agent Teams
openjiuwen/agent_teams
Context and State
- Context Engine
openjiuwen/core/context_engine/context_engine.py - Session State
openjiuwen/core/session - Checkpoint Recovery
openjiuwen/extensions/checkpointer/redis/storage.py
Development Capabilities
- Prompt Development
openjiuwen/dev_tools - Prompt Optimization
openjiuwen/dev_tools/tune/optimizer/prompt_search/service.py - Observability
openjiuwen/extensions/observability
Other components
- Developer
- User
- Model Provider
Connections 17
- Developer → Agent Interfaces: creates agent
README.md - User → Agent Runtime: submits request
README.md - Agent Interfaces → Workflow Orchestration: configures workflows
README.md - Agent Runtime → ReAct Controller: runs agent
README.md - ReAct Controller → Workflow Orchestration: executes tasks
openjiuwen/core/application/llm_agent/llm_controller.py - ReAct Controller → Context Engine: creates context
openjiuwen/core/application/llm_agent/llm_controller.py - ReAct Controller → LLM Invocation: plans tasks
openjiuwen/core/application/llm_agent/llm_controller.py - Workflow Orchestration → LLM Invocation: invokes model
README.md - Workflow Orchestration → Tool Calling: calls tools
README.md - Agent Runtime → Streaming Execution: streams execution
README.md - Context Engine → Session State: saves context
openjiuwen/core/context_engine/context_engine.py - Session State → Checkpoint Recovery: recovers state
README.md - Agent Interfaces → Multi-Agent Teams: supports collaboration
README.md - Developer → Prompt Development: develops prompts
README.md - Prompt Development → Prompt Optimization: optimizes prompts
- LLM Invocation → Model Provider: uses provider
- Agent Runtime → Observability: traces execution