dsh-teams-x
Run the following command in DeepSeek Harness:
dsh plugin install Mlte0907/dsh-teams-x
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install Mlte0907/dsh-teams-x inside DeepSeek Harness to install this plugin; the full source is available at https://github.com/Mlte0907/dsh-teams-x
About this plugin
When a DeepSeek Harness session must carry multi-step, multi-role engineering work, a single-agent setup quickly runs out of room: task dependencies go untracked, member states stay invisible, and failures leave no clear owner. dsh-teams-x solves exactly that by turning one session into a coordinated, auditable multi-agent team.
Under the hood, a captain decomposes the goal into a dependency-aware task DAG, spawns durable member subagents, and orchestrates them through persistent mailbox messages. The Web UI renders a live all-SVG activity panel so you can watch the team work in real time. Six quality gates (requirements, implementation, verification, review, repair, integration) enforce completion rules, and failed tasks automatically derive repair rounds capped at three by default. From v0.2 onward, profile-based team templates enable one-click setup, an in-browser Staged plan editor supports row-level batch revisions, in-session team cards collapse event streams into readable summaries, and a slash command invokes the panel on demand.
If you run long-horizon engineering flows on DeepSeek Harness 0.1.5-rc.2—full-stack scaffolds, cross-repo refactors, CI debugging—and want an explicit task graph, clear role assignments, and auditable coordination records, the thirteen teamsx_* tools plus the all-SVG panel turn a black-box chat into a team you can actually see. The compatibility shim is isolated in a single src/compat.ts file, so host upgrades rarely ripple beyond that one change.
Screenshots
Use Cases
- Multi-role, dependency-aware task orchestration in long-horizon engineering flows
- Quality-gate validation and automatic repair loops for CI debugging or cross-repo refactors
- Transparent in-session visibility of member states, task progress, and coordination records
Best For
- Developers running complex engineering flows on DeepSeek Harness
- Teams that need auditable multi-agent coordination records and explicit task DAGs
- Engineers who prefer persistent-mailbox coordination and quality gates over black-box chat
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.