Run method-guided coding sessions with Loushang
Use Loushang to run recoverable coding-agent sessions with model routing, durable context, tool governance, methods, stages, and traceable delivery artifacts.
npx skills add agentskillexchange/skills --skill run-method-guided-coding-sessions-with-loushang
Use Loushang when an operator wants coding-agent work to follow a structured method instead of an ad hoc prompt thread. Upstream positions loushang code as a CLI and terminal workbench for software development with model routing, persistent sessions, tools, extensions, and method-guided delivery.
What this skill actually does
The repeatable workflow is concrete: install from source, list configured models and commands, start a session from a repo intent, let Loushang keep the method, roles, stages, tools, session record, and work products as runtime objects, then inspect or resume the delivery record. Invoke it instead of using a model chat or coding agent normally when the work needs recoverable state, explicit method constraints, governed tool use, and auditable acceptance expectations. The scope boundary is method-guided coding sessions and delivery traces, not a generic LLM gateway, Python framework, or model SDK listing.
Inputs and prerequisites: Python, uv, git, configured model providers, and the Loushang CLI..
Setup notes: Clone the repository, create a virtual environment with uv venv .venv , activate it, then run uv pip install -e “.[dev]” . Check setup with loushang –help , loushang –list-models , and loushang –list-commands .
Source and verification boundary: use https://github.com/zhnt/loushang as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Custom Agents workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.