AI Agent Hub
Back to plugins
⚙️

dsh-clawo

Workflow Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Enderfga/dsh-clawo

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

In DeepSeek Harness, run dsh plugin install Enderfga/dsh-clawo to install; the plugin source is at https://github.com/Enderfga/dsh-clawo

About this plugin

dsh already ships subagent-claude-code and subagent-codex for delegating to a single coding CLI, but that leaves a gap when a task genuinely calls for multi-engine collaboration. dsh-clawo closes that gap: it is a zero-dependency YAML bundle that registers Claw Orchestrator as an ACP subagent provider, so one subagent call can either let a single engine answer outright or convene a council of engines that debate in isolated git worktrees and hand back a synthesized result.

Four operating modes cover the main workflows. The default single mode routes the prompt to one engine. Council mode spins up several engines in separate worktrees, lets them debate, and returns a consensus answer. Ultraplan is a long-horizon planning pass for complex, multi-step tasks. Ultrareview dispatches parallel reviewers across the working tree for a coordinated sweep. Selecting a mode is as simple as prefixing the prompt with a slash command, such as /council fix the failing test.

This bundle suits dsh users who have already validated single-engine subagents and want to go further: cross-model code review, long-horizon architecture planning, or multi-pass verification before a release. A practical caveat is that council is a heavyweight operation — a measured two-round run took roughly nine minutes — so it is best treated as a deliberate checkpoint rather than a default step.

Use Cases

  • Cross-validating critical code changes with multiple models
  • Long-horizon planning and decomposition before major refactors
  • Parallel multi-reviewer sweep of the working tree before merge

Best For

  • Developers already using single-engine subagents in dsh
  • Code teams that benefit from cross-model collaboration
  • Maintainers who value multi-pass code review quality