dsh-kanban-flow
Run the following command in DeepSeek Harness:
dsh plugin install thomasvvugt/dsh-kanban-flow
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install thomasvvugt/dsh-kanban-flow in your DeepSeek Harness terminal to install the plugin; the full source code is available at https://github.com/thomasvvugt/dsh-kanban-flow
About this plugin
Working alongside an agent in DeepSeek Harness, tasks tend to scatter across conversations: who owns what, how far progress has come, who has the final say — all left to verbal agreement. dsh-kanban-flow turns that fuzzy collaboration into a fixed pipeline. You drag a card from Backlog to To Do; the agent picks it up, works on it, narrates every step in its own task session, and parks it in Review when it needs your input.
The backbone is a five-column workflow (Backlog, To Do, In Progress, Review, Done) with every legal move enforced at the tool level: agents cannot grab unassigned work, cannot skip Review, and cannot close items on their own — illegal moves are rejected with a clear explanation. Each card spawns a dedicated agent session named after the item where progress narration lives; hovering a card reveals the agent's latest status note from its most recent turn; replying on a Review card sends the item straight back to In Progress in the same turn. Enable Require Confirmation per board and the agent can never reach Done without your drag.
Built for developers who pair with agents inside DeepSeek Harness but want to keep the final call. Board data is stored as a local, schema-versioned JSON file per workspace — no external services, no remote calls, no build step at install time.
Screenshots
Use Cases
- Manage agent tasks end-to-end on a fixed five-column board from assignment through human review
- Enforce that agents cannot skip the Review column or self-close items without explicit human approval
- Track per-item agent progress in dedicated task sessions with step-by-step narration and status notes
Best For
- Engineers pairing with agents in DeepSeek Harness who want a controllable, rule-enforced workflow
- Teams that need strict human-agent role separation and mandatory review gates for multi-task pipelines
- Independent developers who value local-first data storage with zero external service dependencies
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.