dsh-ha-orchestrator
Run the following command in DeepSeek Harness:
dsh plugin install Saktawdi/dsh-ha-orchestrator
Paste the following prompt into your AI chat to install this plugin:
You can quickly deploy and enable this extension in DeepSeek Harness by running the installation command, and its complete open-source repository address is https://github.com/Saktawdi/dsh-ha-orchestrator .
About this plugin
When long-running tasks are suddenly interrupted by model glitches or service timeouts, traditional workflows often crash completely, wasting valuable context and compute. HA Orchestrator steps in as a resilient task manager that automatically detects call failures and seamlessly switches to backup models for retries. Coupled with intelligent cooldown periods, provider-level circuit breakers, and adaptive degradation strategies, it ensures your critical pipelines keep running smoothly. You can finally focus on completing complex assignments without constantly monitoring for transient errors.
Beyond simple failover, this plugin empowers you with an auto-triggered orchestrate tool designed to handle intricate, multi-step operations. When prompted, the model will intelligently decompose tasks into parallel fanouts, sequential pipelines, or supervised review loops using five distinct orchestration patterns. You retain full control to define custom sub-agents with specific models, prompt guides, and tool whitelists or blacklists. Everything is accompanied by real-time progress cards, detailed run histories, and diagnostic slash commands, turning opaque multi-agent chains into transparent, auditable workflows.
This tool is particularly well-suited for deep-dive research projects, large-scale codebase analysis, batch document reviews, and comparative planning across multiple solutions. If you rely on robust, long-horizon AI agents or want to automate the tedious parts of task decomposition and cross-verification, HA Orchestrator provides the reliable execution backbone and trackable output delivery you need to scale your automation safely.
Screenshots
Use Cases
- Seamlessly recover long-running research tasks when models time out by automatically switching to backup instances.
- Break down complex project analyses into parallel sub-tasks for efficient processing and consolidated reporting.
- Enhance batch code review quality through a supervisor pattern that cross-validates multiple agent outputs.
Best For
- Researchers and analysts conducting deep investigations that require uninterrupted, long-horizon model execution.
- Engineers building robust automation pipelines that demand fault tolerance and dynamic task routing.
- Project managers overseeing multi-step decision workflows who want to minimize manual oversight and error recovery.
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.

