Orvain UI

Agent findings

Reviewing what an agent found, and fixing or dismissing each item.

Get Pro to installPro18 blocksReact, Vue and Figma
MCPConnect the MCP server
Launch check · Checkout v264 checks · 1 critical · 2 should fix · 2 note
1:12
Findings5
  1. Empty cart has no Figma frameNoteDesign QACart · Empty state
    47s
  2. Retry test is flaky, failed 1 of 5 runsShould fixTestsretry.test.ts:41
    56s
  3. Refund page still says 14 daysNoteDocsHelp center · Refunds
    1m
  4. 1m

    The handler logs every request header, and Stripe's signing secret is one of them. Anyone who can read the logs could fake a payment.

    1. 87const event = verify(req, secret)
    2. 88logger.info('webhook', req.headers)
    3. 88logger.info('webhook', { id: event.id, type: event.type })
  5. Pay button radius is 8px, Figma says 12Should fixDesign QACheckout · Pay button
    1m

More in Agents#

Agent monitor

Watching one agent run, and giving it the OK when it asks

Agent plan

Reading what an agent will do before it starts, and watching it work through the plan

Agent checkpoints

Going back to an earlier point in a run and trying again from there

Agent findings

Reviewing what an agent found, and fixing or dismissing each item

Agent fleet

Keeping track of many agents at once and answering the ones that ask

Agent graph

Seeing which agents hand work to which, and through which apps

Agent cursors

Editing a document alongside agents and taking or leaving what they change

Agent lanes

Comparing what several agents did over the same stretch of time

Agent queue

Watching a queue of agent jobs work through a pool of workers

Agent trail

A small live card for one agent while you do something else

Agent recap

Catching up on what agents did overnight

Trace waterfall

Finding where a run spent its time across models, tools and calls

Cost waterfall

Seeing where a run's money went, step by step, against a budget

Agent inbox

One place to answer everything agents are waiting on

Approval gates

Getting a run signed off by the people who have to say yes before it goes on

Autonomy dial

Deciding how much an agent may do on its own, app by app

Agent steering

Changing what a running agent does without stopping it

Parallel attempts

Letting several models try the same task and keeping the best one

When to use#

  • An agent that acts in other people's apps, where some of its actions need a yes before they happen.
  • Several agents or several attempts running at once, where someone has to see which is stuck and which is done.
  • Runs long enough that people leave and come back, and need to catch up on what happened and what waits on them.
  • Spend or time that has to be explained afterwards, step by step, against a budget.

The whole set, with its codePro

Pro opens the Code tab on every block here, in React and Vue, plus the Figma file and the MCP server for your editor.

Open in BlocksSee plans