dsh-pouch
Run the following command in DeepSeek Harness:
dsh plugin install moon16u/dsh-pouch
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install moon16u/dsh-pouch in your DeepSeek Harness terminal to install this plugin pack with one command; the full source is available at https://github.com/moon16u/dsh-pouch .
About this plugin
Running day-to-day agent workflows on DeepSeek Harness exposes a steady stream of small friction points: restarting after a config change means hunting for the process in an external terminal, a session left open past midnight keeps reasoning from yesterday, grabbing a Session ID requires digging through URLs or console logs, gateways that authenticate by User-Agent are literally unreachable with stock settings, adding an MCP server means editing YAML and restarting, and the Models page 404s against any gateway whose model list lives at a non-standard path. dsh-pouch bundles all of these into six pocket-ready plugins that land with a single dsh plugin install—no hand-written config files required.
The capabilities span three layers. Session infrastructure: a safe detached restart with a 3-second grace period, per-turn wall-clock and timezone injection so long sessions never drift, and a one-click Session ID badge in the web header. Capability extension: real-time web search backed by the Tavily REST API, and full control over LLM request headers—including overriding the User-Agent that DSH reserves for itself, with environment-variable interpolation and per-model overrides, plus a dedicated settings page. Runtime management: a full MCP server console with hot add/edit/enable/disable, per-tool switches, live SSE status, and mcpServers JSON import/export; and a model-listing adapter where a single mapping rule aligns any gateway's non-standard endpoint and field shape to what the Models page expects.
Built for developers who run DSH as a daily agent pipeline, integrate non-OpenAI gateways, or want MCP servers under visual management. Every plugin mounts as an independent fiber, leaves no trace when removed, and does not alter DSH's core architecture.
Screenshots
Use Cases
- Safely restarting DSH from within a live agent session after editing plugins or configs
- Integrating non-standard gateways that require custom request headers or a non-OpenAI model-listing endpoint
- Dynamically adding, toggling, and monitoring MCP servers at runtime without restarting DSH
Best For
- Individual developers or small teams running DSH as their daily agent runtime
- Platform engineers integrating multiple non-OpenAI gateways with custom header and endpoint needs
- AI infrastructure engineers seeking visual, runtime MCP server management
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.