Preview Before You Approve

Simulated. No real agent runs, no real browser is involved, and no email is sent.

Experiment

Firefox with agent sidebar (simulated)
How the host builds the preview
The host builds the approval preview from tool metadata Four lanes: the site, the agent, the host, and the person. In five steps the site lists the effects of a tool, the agent calls the tool, the host shows a preview built from that list, the person decides, and the host runs the tool only after Allow. Site Agent Host Person 1 Tool lists its effects 2 Agent calls the tool 3 Host shows preview 4 Person decides 5 Host runs the tool after Allow

Try this

  1. Pick style A. Press Allow. Answer the quiz.
  2. Pick style C. Press Play on the timeline. Press Allow. Answer the quiz again.
  3. Watch the live score and compare the bars.

Pain point

People approve agent actions without knowing what will change.

Problem statement

Approval prompts say little. People press Allow without knowing who gets the email or what is shared. Nobody checks whether they understood.

What sparked it

Mozilla's W3C WebExtensions group notes (issue 1041) say no browser primitive lets a user see, scope, or withdraw an agent's access. WebMCP lets a tool call requestUserInteraction() to ask the user.

Why this approach

The preview shows what will change, in plain words, before the click. A short quiz then checks understanding.

Hypothesis tested

We believed that an exact preview improves understanding. If it holds, style C will score higher than style A.

How to try it

  1. Open the Prototype tab. Press Start a new round.
  2. Read the chat. Press Allow or Deny.
  3. Answer the three questions.
  4. Repeat with another style and compare the bars.

What this does not do

Notes

Traces

Prompts used to make this prototype
StepPromptModelToolDate

Cost

Cost to make this prototype
StepModelInput tokensOutput tokensCost (USD)
Total

Measured: the author runs used 191,740 tokens in total. Estimated: the split into input and output tokens (90 percent input, 10 percent output) and the dollar cost, at $2 input and $10 output per million tokens for Sonnet 5.5 (about $0.54). The run reported one total only. The lead session's planning, research, and publishing are not included. The prompts in the Traces section are listed in the order they were given.