AI Agent Hub
Back to plugins
dsh-ariadne preview

dsh-ariadne

Workflow Updated 2026.09.04

Run the following command in DeepSeek Harness:

dsh plugin install Zayzz-pixel/dsh-ariadne

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Zayzz-pixel/dsh-ariadne in your terminal to install the plugin from https://github.com/Zayzz-pixel/dsh-ariadne; after installation the Ariadne tab appears at the top of each new session.

About this plugin

Multi-turn discussions tend to sprawl: directions expand, trade-offs blur, and by the end the actionable items are buried in context. Ariadne is a visual decision workbench for DeepSeek Harness that externalises the directions, sub-branches, and parked threads from a divergent conversation into a draggable, focusable, collapsible session map, keeping exploration broad while giving the final choice one clear execution path.

On the map you can focus any node to review the current understanding, open questions, and next step; pull promising directions into a decision pool; and have the Agent generate an explicit Task / Decision / Checkpoint execution graph from that pool. You then trigger execution node by node, with results, evidence, and artifact references written back into the same Run. After a page refresh the full state is restored from persistent session events, with no re-conversation needed.

It is built for teams and solo developers who use DeepSeek Harness for structured decision-making, option comparison, and execution planning, anyone who wants to turn a long conversation into a reviewable, resumable workflow. The current release targets DSH 0.1.1-rc.2 Web profile on Node.js 18+.

Screenshots

Use Cases

  • Turn a sprawling multi-turn discussion into a reviewable, focusable directional map
  • Convert confirmed directions into an explicit Task / Decision / Checkpoint graph and advance node by node
  • Restore work state across sessions with full evidence and artifact references

Best For

  • Users doing structured decision-making and option comparison in DeepSeek Harness
  • Teams that need to turn long conversations into actionable, resumable workflows
  • Developers who value traceable, reviewable, interruptible execution