English | 中文

Loushang is a method-native AI work system for running complex work from intent to verified delivery.

Current focus: loushang code, a CLI and terminal workbench for software development with model routing, persistent sessions, tools, extensions, and method-guided delivery.

Modern AI agents can plan and act, but complex work still breaks down when context is lost, execution cannot be resumed, tools are hard to govern, and results are not verified.

Loushang treats methods, stages, roles, tools, sessions, and work products as runtime objects. The goal is not just to make agents smarter, but to make complex work more reliable, recoverable, auditable, and deliverable.

• loushang code: a coding-focused CLI and terminal workbench.
• loushang.ai: a provider-aware AI SDK with model registry, streaming, tool calls, and cost helpers.
• Sessions: persistent coding sessions with resume, fork, export, and diagnostics.
• Tools: built-in coding tools and configurable tool surfaces.
• Extensions: project-level extension hooks, custom tools, dynamic resources, and commands.
• Methods and skills: method-guided coding turns and reusable workflow assets.

Loushang is in early development. The recommended path is to run it from source.

git clone https://github.com/zhnt/loushang.git
cd loushang

uv venv .venv
source .venv/bin/activate
uv pip install -e ".[dev]"

loushang --help
loushang --list-models
loushang --list-commands
loushang -p "Inspect this repository and summarize what it does."

You can also run make bootstrap, which creates .venv/ with uv and installs the project in editable development mode. The Makefile does not currently provide a make install target; use make bootstrap for local development or make install-binary for a local binary install.

For local development in this repository, use the project virtual environment in .venv/.

• Method: a structured work contract that defines roles, phases, workflow, constraints, artifacts, and acceptance expectations for a class of work.
• Session: a durable coding conversation and execution record that can be resumed, forked, exported, and inspected.
• Tool: an executable capability made available to the agent under policy.
• Extension: project-level Python code that can contribute hooks, tools, resources, commands, and flags.
• Model provider: a concrete AI provider endpoint and model resolved through the model catalog.

• Documentation Home
• Getting Started
• User Guide
• Concepts
• AI SDK
• Examples
• Reference

• Coding examples show CLI/session/tool/extension scenarios.
• AI SDK examples show model lookup, complete, stream, tools, and typed contexts.

• V1: loushang code as the primary product surface for software development work.
• V2: loushang work as a personal complex-work workbench, with code, research, and ppt as specialized flows.
• V3: daemon, method market, and model gateway foundations.
• V4: team workflows, shared runs, approvals, budgets, and audit.
• V5: managed runtime for method-bound complex work.

Loushang is in active early development.

The current stable focus is loushang code and the underlying loushang.ai SDK. Broader work surfaces such as loushang work, loushang research, and loushang ppt are part of the roadmap and should be treated as evolving product directions.

Loushang was initiated by Heng Zhou. He has long worked across low-code systems, workflows, databases, model-driven engineering, DSLs, architecture methods, systems engineering, and artificial intelligence, with a focus on operationalizing ontology and methodology into infrastructure for complex-work delivery.

For questions, feedback, collaboration, or a community group invitation, contact: zhnt@foxmail.com.

Loushang learns from public design and engineering patterns in projects such as OpenAI Codex, pi, python-prompt-toolkit, browser-use, Kimi CLI, superpowers, gstack, openclaw, and hermes-agent. These projects are references and inspiration; unless listed in THIRD_PARTY_NOTICES.md, this repository does not include or redistribute their code.

Loushang is licensed under the Apache License 2.0 unless a file states otherwise.

When redistributing source code, binaries, documents, or modified versions, keep LICENSE and NOTICE, and retain attribution in product documentation, About/Credits pages, or equivalent third-party notices.

Third-party dependency information is available in THIRD_PARTY_NOTICES.md.