01 / 05
Hold the objective
Something has to own the goal for longer than one agent run, and survive the process that started it.
Skyflo's answer
A mission: one objective with state that outlives the session.
Definition first
What is an engineering harness?
The layer around AI coding agents that holds one objective together: what it has to cover, and how Skyflo covers it.
Coding agents write code. An engineering harness is the layer around them that holds one objective together across repositories, agent runs, review, and memory.
category / system boundaryobjective held
Engineering harnesspersistent boundary
01 · intent
Objective + plan
02 · boundary
Scope + tools + surfaces
execution unitAI coding agentwrites code
03 · evidence
Independent review
04 · continuity
Source-linked memory
The agent operates inside the boundary.The boundary survives the run.
Definition
The layer around the agent.
An engineering harness is the layer around AI coding agents that holds an engineering objective together: the plan the work runs against, the delegation of bounded pieces to workers, the surfaces the work runs on, the review of what came back, and the context that carries to whatever comes next.
The agent writes the code. The harness is everything that has to be true for the code to be the right code, in the right places, for a reason someone can still recover next quarter.
Writing code is only the middle of the system.
Models are capable enough that coordination, independent review, and continuity are often the limiting system problems. Engineering harness is one name for the layer that handles them.
work / unresolved contextoutside the function
01
Which repositories does this change touch?
02
What shape did we approve?
03
Who checks the integrated result?
04
What survives when the window closes?
Better model output does not answer a missing system boundary.
For a few years the interesting question was how well a model could write a function. Models and coding agents have improved quickly, but capability still varies by task, repository, provider, and the evidence available to the agent.
What did not get solved is everything around the function. Those are not model problems. Adding a better model to a system with no answer to them produces the same result faster.
The word is not Skyflo's invention. “Harness” is an established product noun elsewhere, and harness engineering is being shaped in public by several groups at once. It describes a layer of coordination for engineering work, one application of Skyflo's company-wide operating system.
What a harness has to hold.
Any system that answers all five is a harness, whatever it calls itself. Each card begins with the general responsibility, then shows how Skyflo answers it.
02 / 05
Decide before doing
An approved boundary, established once, that every worker inherits instead of re-deriving.
Skyflo's answer
Plan mode waits for your approval before anything changes, and each mission's access mode sets what agents may do without asking.
03 / 05
Bound the workers
Delegated work needs a scope, a toolset, and somewhere to write that stays apart from everyone else's work in progress.
Skyflo's answer
In Orchestrate, each coder specialist works in its own git worktree from a packet that names its scope. Outside Full access, external agents run inside a macOS sandbox profile.
04 / 05
Check the result independently
The thing that assesses the work should not be the thing that produced it, and should not be able to change it.
Skyflo's answer
A separate reviewer that can read, search, ask and report, and cannot edit a file. Orchestrate adds one by strategy and risk.
05 / 05
Keep what was learned
What one piece of work discovered should be retrievable by the next, with the work that produced it attached.
Skyflo's answer
Memory learned from finished missions under evidence rules, source-linked and removable. A command sequence verified in two repositories can become a Skill.
Skyflo turns the boundary into a running system.
The category is general. Skyflo's implementation stacks surfaces, a persistent mission, bounded specialists, typed capabilities, and memory learned from the work around one objective.
harness / cutawayexploded view
Surfaces. Desktop, browser, terminal; all on one mission.
SF-05 · one mission, several surfacesInteractive cutaway · hover a layer to inspectOne objective passes through all five
What it is not.
The word is new enough that it gets attached to adjacent categories. These three come up most.
adjacent · 01
Not an AI coding agent
A coding agent executes development work. An engineering harness owns the objective, boundaries, evidence, review, and continuity around that work.
adjacent · 02
Not an IDE or an editor
The editor is one surface the work happens on. So is the terminal, the browser, and whatever runs on a schedule. A harness is defined by what it holds across those surfaces, not by which one it lives in.
adjacent · 03
Not an orchestration framework
A framework gives you primitives for building coordination. A harness is a product that applies those primitives to a running workflow, including its approval, isolation, and review boundaries.
When the harness learns from its own work, it becomes a HyperAgent.
A harness holds one objective together. Nothing in the definition makes it better at holding the next one. Skyflo improves how it improves.
Since Desktop 1.1.0, Skyflo learns from the missions it completes and carries what they teach into later work. A command sequence verified in two repositories can become a reusable Skill, and Skyflo can change the policy it uses to route Orchestrate missions between agents and models.
The harness boundary is what makes that safe. The part that proposes an improvement never judges it. Approvals, credentials, spending limits, and kill switches sit outside anything Skyflo can change about itself, and any change to what it learned or how it routes can be reverted, which changes what it does next, not work already done.
Skyflo is the agentic operating system for your company. Missions hold goals, access and the record across agent work, from research and analysis to software delivery. For engineering missions, the harness adds repository coordination, isolated worktrees and independent code review. It is one application of the platform, with the same access rules underneath.
See one mission run.
Skyflo Desktop runs missions on your Mac, against your repositories, with the coding agents you already use or your own model key.
Start free. macOS 14+, Apple silicon and Intel. Agent subscriptions and model keys you bring are billed by their providers.