Clarity Harness · Open Source
Compound your context.
Every agent run makes the next one better. Incubate your agents like new hires — check their work, give verdicts, bank every correction — until they earn autopilot. Then share your agents and harnesses across the team. Your subjective taste is the advantage. The harness compounds it, right in the flow of work.
1,437 builders already in line · free & open source · no license keys
The Clarity Harness — mission control for your agents.
The pain
Agents made you fast.
They didn't make you sure.
You bought agents. You got drafts.
Claude Code, Cursor, your own stack — throughput 10x’d overnight. But every artifact is a draft until someone you trust says it isn’t. You didn’t remove the bottleneck. You moved it.
You’re the eval now.
There are evals for models. There are none for your agents’ actual work-product — the stories, specs, and PRs they ship into your backlog. So your senior people read everything. Or worse: they stop reading.
Your corrections don’t compound.
You catch the same mistake every week. The fix dies in a Slack thread. Tomorrow’s run starts with zero memory of being wrong. All that judgment, spent — none of it banked.
Speed without compounding is just expensive slop.
The promise
Every verdict compounds.
The harness closes the loop: capture the run, deploy the agent, review with a verdict, learn from the correction. Today's annotation is tomorrow's default. Week one feels like review. Week six feels like leverage.
Capture
Every run recorded, receipts attached
Deploy
The agent runs your workflow
Review
You give the verdict in minutes
Learn
Corrections banked into the next run
Capture
Every run recorded, receipts attached
Deploy
The agent runs your workflow
Review
You give the verdict in minutes
Learn
Corrections banked into the next run
Proof · every run shows receipts
Every artifact shows its chain.
There are agent runs, and one question that matters: did the input produce the expected output — and how does it get better next time? The harness makes that question answerable.
01
Quoted from here
Every requirement links back to the exact transcript line it came from. No mystery provenance.
02
The chain that got you here
See which agents worked together and in what order — story writer with requirements synthesizer, on this run, from this input.
03
The diff from before
What did this version add? What changed from the last one? The full artifact, one click away.
04
Your verdict
Approve it — or annotate it: “this requirement was too narrow, specify broader.” Save. Regenerate. The harness learns your judgment.
Proof · we run on it
This is how we're evolving our work.
If you're a product manager, you're going to have a story writer at the top — and a stack of agents underneath it doing the work you used to do by hand. The screenshots on this page are our production instance, not a demo: 8 agents, 129 runs, 155 verdicts in the queue — the harness running the company that built it.
◆ agent
Story Writer
at the top
Turns captured requirements into stories, drafted and waiting for your verdict — not your keystrokes.
◆ agent
Clarity Orchestrator
routing the work
Orchestrator calls the agent, agent calls the skill. Point your coding agent at it and the whole stack composes.
◆ agent
Interview Agent
capturing input
Sits on your meetings and conversations, turning what people actually said into structured evidence.
◆ agent
Requirements Synthesizer
from transcript to spec
Works with the story writer to pull requirements straight out of transcripts — quoted, sourced, traceable.
One philosophy, every scale
Individual. Domain. Company-wide.
Individual
Your own harness
Every individual has their own harness — your agents, your runs, your judgment, your data. In the flow of work: writing stories, making prototypes, agents doing stuff.
Domain
Domain-specific harnesses
Privacy, payroll, compliance — wherever tacit knowledge lives, a domain harness turns the expert’s judgment into something every run can use.
Company-wide
The company brain
Department and company-wide harnesses share one brain across your people and agents. That’s the layer we build with design partners today.
See the company brain →What ships day one
Everything we run internally. All of it.
The full harness UI
Mission control, run views, judgment queue — the exact interface in the screenshots above.
Run capture with full provenance
Every run recorded: source input, agent chain, tool calls, decisions taken and not taken.
The verdict system
Approve, accept with edit, reject, defer — with annotations that persist as structured data, not Slack threads.
Versioned datasets
Your corrections become a dataset with lineage — dev, staging, prod — that regenerates your agents better.
MCP serve
Point Claude Code or any coding agent at your harness — orchestrator calls the agent, agent calls the skill.
Our internal setup guide
The same playbook we used to stand up the harness we run Clarity on. First harness running the same afternoon.
Built in the open
A community for builders who believe agents deserve better than vibes.
The harness ships free and open source because the thesis is bigger than a tool. This is the Clarity / Epistemic Me bet — and we're building a community of AI builders who share it:
◆
Subjectivity is the missing primitive
Autonomy doesn’t come from bigger models. It comes from knowing whose goals, whose judgment, and whose context a run answers to.
◆
Alignment is layered
Individual, team, and organization goals aren’t the same thing — a real harness reconciles all three instead of pretending one prompt covers it.
◆
Closed loops beat bigger prompts
True autonomous agents are grown, not prompted: run → verdict → learn → run again, with the human judgment banked every cycle.
If that's your thesis too, don't just star the repo — come build the loop with us.
Open source · early access
Start compounding before everyone else.
Early access gets the repo first, plus the setup guide we use internally — first harness running the same afternoon. Open source, no license keys.
1,437 builders already in line