dsh-farm-dispatch
Run the following command in DeepSeek Harness:
dsh plugin install sangning-h/dsh-farm-dispatch
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install sangning-h/dsh-farm-dispatch in DeepSeek Harness to install this plugin (source: https://github.com/sangning-h/dsh-farm-dispatch).
About this plugin
DSH supports multiple presets running in parallel, yet sub-agent dispatch discipline often lives in scattered personas or verbal conventions, causing inconsistent behavior, redundant dispatch, and uncontrolled granularity. dsh-farm-dispatch codifies a complete dispatch protocol task allocation, thrift, execution style, traceability, and async acceptance into a global AGENTS.md file that DSH's native agent-instructions mechanism injects as a user-role instruction, so every preset with sub-agent capability operates under one shared rule set.
The plugin also ships a cordis-async preset (creative mode plus async dispatch discipline) and idempotently syncs it to the roster at startup. The full protocol text lives in a single file at $DSH_HOME/AGENTS.md; editing it once propagates to all presets and all active conversations on the next turn, enforcing a true single source of truth.
Ideal for individual developers and small teams who heavily rely on multi-preset workflows and need a stable, auditable sub-agent dispatch rhythm. If you want dispatch behavior to be traceable, reusable, and no longer dictated per session, this plugin bakes the rules into the infrastructure layer.
Use Cases
- Unify sub-agent dispatch behavior across multiple active DSH presets
- Edit one protocol file and every preset inherits the change on the next turn
- Maintain background dispatch and two-phase acceptance discipline in async creative mode
Best For
- Individual developers who rely heavily on multi-preset DSH workflows
- Small teams that need a consistent sub-agent dispatch rhythm
- Automation enthusiasts who want workflow rules baked into the infrastructure layer
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.