dsh-init-command

Every project deserves an AGENTS.md that tells AI agents how to build, test, and follow the code style. Writing one by hand is time-consuming and easy to get wrong, especially when the project is still a fresh skeleton. dsh-init-command adds an /init slash command to DeepSeek Harness that lets an LLM do the heavy lifting for you.

The workflow is two-phase. Phase one collects the first two levels of the directory tree and asks the model to identify the project type, language, framework, and build toolchain, returning a structured classification. Phase two embeds that result plus the directory structure into the generation prompt and produces a full AGENTS.md covering sections like Project Overview, Build and Test Commands, Code Style Guidelines, Testing Instructions, Security Considerations, and AI Agent Guidelines, writing the file directly to disk. The classification phase runs with thinking off by default to save tokens; the generation phase streams the model reasoning live into the session while the final text goes straight to the file, keeping the conversation clean. A built-in rule in the AI Agent Guidelines section also reminds the agent to update AGENTS.md after every completed task, so the document stays in sync as the project evolves.

Beyond the core generation, the plugin handles common cold-start chores: git init with a main default branch, downloading a .gitignore matched to the project type, and creating an initial commit. When the directory is empty it asks the user for a project description instead of guessing; an existing AGENTS.md is overwritten with the old content fed to the model as a rewrite reference. Zero dependencies, no build step required. It fits developers who use DSH as their daily coding environment and want AI agents to understand project conventions fast, whether bootstrapping a new repo or picking up a documentation-less codebase.