One grade for agent readiness
ora runs real agents through your product and hands back a single score — how far they get, where they stall, and what it costs them to succeed.
agentfront.sh — by ora
Agents are your newest users. Greet them at the door.
agentfront runs real AI agents through your real product, replays every step of their journey, and grades how ready you are — so the fastest-growing users on the internet don't bounce off your front door.
now onboarding from the waitlist · small weekly batches
ora · live runstreaming
ora run yourstore.com --journey checkout
A−
claude-code · boot harnessGET / → 200 · 214msthink: "find pricing"click nav "Pricing"GET /pricing → 200stall: cookie bannerfill #emailPOST /signup → 201journey checkout ✓trace: 42 spansreadiness A−insight: label the buy buttonre-run → A
product
See your product the way an agent does.
Not a lab simulation — real harnesses running real journeys against the product you already ship.
Trajectory replay
Every run becomes a living graph: each click, call, retry and dead end, laid out left to right so you can read an agent's journey like a sentence.
Real harnesses, not simulations
Claude Code and the harnesses your users actually bring. ora drives them at full capability against your real product — no mocked browser, no synthetic DOM.
Insights that name the fix
Not just "the agent failed" — which step it gave up on, why, and the change that unblocks it. Fix, re-run, and watch the grade climb.
Live from the first step
Runs stream to your screen in real time as the agent works. No refresh button, no waiting for a report.
A front door, not a side door
Agents shouldn't have to scrape their way in. Ship a first-class agent front and greet the fastest-growing users on the internet properly.
how it works
Point, run, read, fix. Then watch the grade climb.
- 01
Point ora at your product
Give us a URL. No SDK to install, no code to change, nothing to configure.
- 02
Agents run real journeys
Sign-up, checkout, support, search — driven end to end by real agent harnesses.
- 03
Read the trajectory
Every step, span and stall replays as a graph you can walk through, live as it happens.
- 04
Fix, re-run, re-score
Ship the change the insight named, run it again, and watch your readiness grade climb.
from the blog
Field notes from the agent front.
faq
Fair questions.
What is an agent front?
The surface of your product that AI agents see and use. Humans get a front end; agents today get whatever survives scraping. agentfront measures that surface, shows you where agents succeed or stall, and helps you make it first-class.
How is agentfront related to ora?
agentfront.sh is built by ora. Same platform, same team — agentfront is the home of our agent-readiness product: run agents against your product, replay trajectories, and score the result.
Do I need to change my product to try it?
No. ora points real agent harnesses at your existing product from the outside — you get a readiness grade and trajectory replays without installing anything. Changing your product is what you do after, once the insights name the fixes worth shipping.
Which agents and harnesses do you support?
We drive real harnesses — Claude Code among them — at full native capability, across the providers and models your users actually bring. The harness catalog grows continuously; if your users' agent isn't covered, tell us and it moves up the list.
When do I get access?
We're onboarding from the waitlist in small weekly batches so every team gets real attention. Join below — you'll get a confirmation right away and a note from us when your slot opens.
What do you store about my product?
Run traces: the steps, requests and pages the agent touched during a run. Traces stay scoped to your organization, are never used to train models, and you can delete them at any time.
join the waitlist
Be first through the door.
We onboard in small weekly batches so every team gets real attention. Tell us where to knock.