Early development — pre-beta

Features

Everything in Dyson Harness exists for one outcome: getting large-scale work done with a coordinated team of agents instead of one overextended chat.

Orchestrate the work

A team of agents with clear roles, directed by you.

Multi-agent orchestration

Work
Plan
Ask

Coordinate the work across sessions and agents in Work, Plan, or Ask mode, keeping every effort pointed at the same outcome.

Delegated subagents

Drone
Explore

Hand pieces of the work to subagents in Drone or Explore mode. Parallel progress without losing the thread of the whole.

A deliberate change loop

01Plan
02Enhance
03Build
04Review
05Fix

For larger changes, the design is captured in a plan you can review before implementation begins. Fewer surprises, more deliberate results.

Persist and control

Long jobs stay on track — and under your control.

Persistent & resumable sessions

Sessions persist and resume, so ambitious work isn't lost to a restart. Come back later and pick up exactly where things stood.

Context control

Bonsai context trimming steadily compresses and prunes the session context, significantly reducing its footprint over time while keeping long-running sessions clean, focused, and free of accumulated noise.

Per-mode tool limits

Each agent mode gets only the tools it needs. Researchers research; implementers implement — a sensible boundary that keeps work focused.

Models and providers

Strong models, your choice of where they run.

Built for capable models

Built to get the most from strong, modern models.

  • GPT 5.6Terra · Luna
  • GLM5.2 · 4.7
  • KimiK3 · K2.7

OpenAI-compatible providers

Connect the inference provider that fits your workflow, whether it runs in the cloud or on your own infrastructure.

  • Completions
  • Responses

Managed plan providers

Put the subscriptions your team already uses to work, without a separate API account or another billing setup.

  • OpenAI Codex
  • Kimi coding plan

Tools and platforms

The everyday tools real work requires, in one place.

Skills browser showing searchable skills with download actions
Find and install reusable skills without leaving the workbench.

Discover ready-to-use skills, download them in moments, and give every workflow — from research and coding to everyday operations — the tools it needs to move forward.

Supported open-skills providers

  • SkillsHub
  • skills.sh
  • ClawHub
  • Skills Directory

Windows browser control

On Windows, agents can drive a real browser — navigating, reading, and interacting with pages — so web-based work gets done end to end.

Workspace editing, shell & web research

Agents edit files in your workspace, run shell commands, and research on the web — the everyday tools real work requires, in one place.

Custom MCP servers

Connect your own Model Context Protocol servers to give agents organization-specific tools — internal APIs, databases, and services — right alongside the built-in ones.

openrules spec support

Dyson Harness supports the open openrules specification for declaring which project docs and skills agents should load — so your team's conventions travel with the repo. openrules spec

Dynamic MCP call chains

Define multi-step MCP workflows in structured JSON with branching, loops, and tool composition — JDSL turns tool calls into reusable automation that agents can execute on demand.

Subagent event messaging

Subagents can message one another through events as work unfolds — sharing progress, findings, requests, and handoffs without waiting on a central loop. Parallel tasks stay coordinated, responsive, and focused.

Get started