dsh-provider-dispatcher
Run the following command in DeepSeek Harness:
dsh plugin install wangmuy/dsh-provider-dispatcher
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install wangmuy/dsh-provider-dispatcher in a DSH project directory to add the plugin to your profile; the source repository is https://github.com/wangmuy/dsh-provider-dispatcher .
About this plugin
When you have multiple search providers or web-fetch providers installed in your DeepSeek Harness profile, they each register directly into the global ctx with no coordination between them. Whichever loads last silently overrides the earlier ones, leaving you with a single working provider and no way to combine their outputs. dsh-provider-dispatcher solves this by inserting a transparent dispatcher layer between the harness and your child plugins, so they coexist under one unified entry point while the caller only sees a single provider.
The framework is deliberately capability-agnostic: it ships with no built-in opinion about web search, web fetch, LSP, subagents, or anything else. All dispatch logic lives in a setup script that the framework simply calls. The script isolates the target service, creates a recording registry that captures every register* call the children make, mounts the children under a private sandbox, and finally registers a proxy on the global ctx that fans one operation out to all recorded providers. Two example setup scripts are included: web-search-setup supports parallel (run every child, merge all results) and bail (try in order, stop at the first non-empty hit) strategies, while web-fetch-setup offers parallel (race for the fastest success) and bail (sequential fallback). Custom merge functions, tool-name remapping, and failure tolerance are all configurable through the params object, keeping composition logic where it belongs—in your script, not the framework.
This plugin is aimed at two audiences. First, users who already run several search or fetch plugins and want their results merged rather than silently overwritten. Second, DSH developers building a custom capability who need a clean, extensible dispatch layer to orchestrate multiple provider plugins behind a single interface. If a single provider covers your needs, you do not need this dispatcher; but the moment you want more than one to work together, it gives you a structured and composable way to make that happen.
Use Cases
- Merge deduplicated results from multiple search providers in parallel or sequential order
- Race multiple fetch providers and return the fastest successful response
- Author a custom setup script to orchestrate dispatch across any combination of child plugins
Best For
- Users running multiple search or fetch plugins in a DSH profile who want them to work together
- DSH workflow builders requiring provider negotiation, result merging, and failure tolerance
- DSH developers who want to extend a custom capability with a pluggable dispatch 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.