How people use Labports

Your coding agent writes up the session; you keep the good parts. Here’s how it works and what it’s for.

How it works

  1. 1Connect once

    Paste one setup prompt into Claude Code, Codex, Cursor or any MCP agent, and approve it in your browser.

  2. 2Say “report this”

    At the end of a session. The agent writes it up with the prompt, what it tried and the screenshots.

  3. 3Polish and share

    Fix a line, add a picture, set the verdict. Keep it private, or share the good ones.

Use cases

  • Keep a build log

    Every session on a side project, written up while it’s fresh, so nothing you learned gets lost.

    Sayreport this

  • Show what you built

    Turn a weekend build into a page worth sharing: the idea, the result, the screenshots.

    Sayreport this, I want to share it

  • Compare agents and models

    Run the same task through two tools and keep a fair, side-by-side record of how each did.

    Saycompare these two runs

  • Remember how you fixed it

    The cause, the dead ends and the fix, so the next time it breaks takes minutes.

    Sayreport this bug fix

  • Catch up on an old project

    Point your agent at a repo and get the story of what’s there and how it got that way.

    Sayreport on this project

  • Follow a project over time

    Group a project’s reports into a series: a timeline from the first prompt to version 1.0.

    Sayadd this to my series

What’s in a report

Three layers, one under the other. Sharing a report shows the story; the rest is yours to open up or keep.

  1. The storyShared when you share

    What you set out to do, what happened, and the screenshots. Readable by anyone.

  2. Technical detailsYour choice

    The approach, the changes and the numbers, for readers who want the how.

  3. The lab logPrivate by default

    Commands, errors and dead ends. Kept to you, with paths and secrets left out.

Start with your next session

Connect an agent in a minute, then ask it to report.

Create your notebookConnect an agent