Book map
10 chapters + appendix. Read in order, or jump straight to the chapter you need.
Meet DSH: Models, Harnesses, and Agents
Separate the model, the harness, and the Agent.
02First Launch: Desktop, Model, Workspace
Connect a model, pick a workspace, and start your first conversation.
03Your First Task: Say What You Mean
Turn a vague request into a verifiable task.
04Workspaces, Files, and Permissions
Understand workspace boundaries, approvals, and destructive actions.
05Sessions, Context, and Memory
Learn why the Agent forgets, and how to work with context.
06Tools and Skills: The Agent's Hands and Mind
Tools, plans, subagents, and Skills — who does what.
07Modes and Presets: Same Model, Different Ways to Work
Standard, Creator, PTC, and Minimal — when to use each.
08Plugins, Bundles, and Profiles: Giving DSH New Powers
Plugin, Bundle, Profile, and how installation works.
09Creator Mode: From User to Extender
What Creator mode can do, its safety boundary, and a real plugin case study.
10Troubleshooting, Safety, and Next Steps
Common failures, cost control, and a safety checklist.
AAppendix
Glossary, prompt templates, and a one-page cheat sheet.
Cognition
A practical decision checklist distilled from a real project, ready to hand to an Agent.
From CLI to DSH Plugin: A Decision Guide for Better Agent Work
Not an API catalogue, but a sequence of decisions that helps an Agent design, implement, test, and publish a plugin with less rework and fewer errors.
Three reading paths
About this book
An unofficial, practical introduction to DSH for people who are about to let an Agent read files, run tasks, and change real things. It does not try to document every API. It builds the right habits first: use a small workspace, start read-only, plan before executing, and ask for evidence before accepting. This English edition was edited for natural English usage; corrections are welcome.