zhnt/loushang

AI-native agent harness for coding workflows by python: multi-model LLM orchestration, stateful sessions, tool governance, traceable delivery, and provider routing for GPT, Claude, DeepSeek, Qwen, Kimi, GLM, and MiniMax.

What it solves

Loushang addresses the reliability and recoverability of complex AI-driven work. It prevents the common failures of AI agents where context is lost, execution cannot be resumed, and results are not verified, by treating work processes as structured, runnable objects.

How it works

The system operates as a method-native AI work system. It uses "Methods" as structured work contracts that define roles, phases, and acceptance expectations. It manages execution through persistent "Sessions" that can be resumed, forked, or exported, and provides a provider-aware AI SDK (loushang.ai) for model routing and tool calls. The current primary implementation is loushang code, a CLI and terminal workbench specifically for software development.

Who it’s for

It is designed for software developers and knowledge workers who need to perform complex, multi-stage tasks using AI agents while maintaining a strict audit trail and the ability to recover work in progress.

Highlights

  • Method-Guided Delivery: Uses structured contracts to ensure work follows a specific workflow and meets acceptance criteria.
  • Persistent Sessions: Durable execution records that allow users to resume or fork coding conversations.
  • Extensible Architecture: Supports project-level Python extensions for custom tools, hooks, and commands.
  • Provider-Aware SDK: Includes a model registry and cost helpers for managing different AI model endpoints.

Related

  • Project
    rpamis/comet

    A resumable long-running task workflow and skill platform for AI-assisted coding that provides structured methodologies and multi-agent coordination for complex software requirements.

  • Project
    AMAP-ML/LongHorizon-Harness

    A loop engineering framework that turns existing AI agents into long-running computer-use systems by wrapping them in a structured plan-act-verify cycle across GUI and CLI environments.

  • Project
    AgentEra/Agently

    An AI application runtime framework that provides structured outputs, observable actions, and signal-driven workflows to build reliable, production-ready AI service backends.

  • Project
    agentspan-ai/agentspan

    A durable runtime for AI agents that prevents execution loss during process crashes and supports long-running, event-driven workflows with human-in-the-loop approvals.

  • Project
    linghungegeg/Linghun

    A local-first, evidence-driven AI programming terminal that integrates LLMs into real projects with system-level constraints to reduce hallucinations and improve engineering reliability.