AI Agent Hub
Back to plugins
⚙️

dsh-worktree-flow

Workflow Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install TY-ang/dsh-worktree-flow

Paste the following prompt into your AI chat to install this plugin:

Run the install command in your DSH terminal to load the multi-repo feature workspace plugin; the source is hosted at https://github.com/TY-ang/dsh-worktree-flow — restart dsh web after installation.

About this plugin

When your project spans several component repositories—frontend, backend, API—the most common agent-session accident is writing into the wrong directory or landing on the wrong branch. You end up re-declaring paths and branch names in every prompt, and one slip undoes the work.

dsh-worktree-flow introduces the concept of a set: bind multiple repositories under one name, and creating a feature workspace auto-generates a git worktree for each component, all checked out on the same feature branch. Combined with the DSH session sandbox, pre-created branches, and an auto-injected identity note on the first step, writing to the wrong place becomes physically impossible. Configuration lives in the sidebar settings page, with auto-discovery of neighbouring repos, live shared-docs reading, editable branch types, and both project-level and feature-level session guidance injection.

It is built for developers who maintain multiple component repositories and want to work across frontend, backend, and API in a single session—especially teams who are tired of hand-typing paths and branch names into every prompt.

Use Cases

  • A single feature spans frontend, backend, and API repos and must be worked on in one session
  • You do not want to hand-type directory paths and branch names into every prompt and instead want the agent to land on the correct target automatically
  • Multiple features are being developed in parallel and their code directories and branches need physical isolation to avoid cross-contamination

Best For

  • Full-stack developers who maintain multiple repositories spanning frontend, backend, and API
  • Developers using DSH / DeepSeek Harness for AI-assisted coding who want to eliminate repetitive path declarations
  • Teams that manage multiple feature branches in parallel and want one-click workspace creation and archival