dsh-agent-preset-switcher
Run the following command in DeepSeek Harness:
dsh plugin install aefuimn/dsh-agent-preset-switcher
Paste the following prompt into your AI chat to install this plugin:
Install this plugin in DeepSeek Harness by running dsh plugin install aefuimn/dsh-agent-preset-switcher; the full source is at https://github.com/aefuimn/dsh-agent-preset-switcher .
About this plugin
In DSH, an agent preset defines a session tool catalog, system prompt, and capability boundary. Officially, once a session starts the preset is locked: the browser RPC returns a lock error for started sessions because historical tool calls would be incompatible with a new composition. If you need to change the tool set or prompt strategy mid-conversation, you must close and recreate the session, losing all context.
dsh-agent-preset-switcher moves the switch moment from request arrival to the step boundary (after one model request finishes, before the next is assembled), reusing the full official AgentPresets.recompose() mechanism. The entry point is the /mode slash command: /mode list shows the current preset and the full roster, /mode
Ideal for developers who need to adjust tool catalogs, prompts, or capabilities in long-running sessions, or for users who want to quickly experiment with different preset scenarios on the same conversation thread. Subagent sessions cannot be switched (their composition follows the parent), concurrent switches within a session are serialized via a per-session promise chain, and on failure the original composition is preserved and execution continues.
Use Cases
- Switching tool catalogs or prompt strategies mid-conversation without losing context
- Experimenting with different agent presets on the same session thread
- Dynamically adjusting agent capabilities during debugging to verify tool compatibility
Best For
- Engineers running long DSH sessions that require mid-conversation preset changes
- Researchers comparing multiple agent presets in a single thread
- Advanced DSH users building complex multi-tool workflows
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.