Working Surfacelink · 10 Nov 2026
02
A thick stack of printed pages folded shut inside a wooden letter box, with one small handwritten card clipped to the lid that a figure seen from behind is reading, the whole scene drawn with wide empty margins on every side.
Simon Willison’s Weblog

Hide your slop PR descriptions behind <details><summary>

Simon Willison · 9 November 2026

How-to · critique & review · artifacts · Decision records

Read the original
Takeaway

The developer Simon Willison opens each AI agent's pull request with a short summary checked by a person, and folds the agent's long description out of sight.

Summary

Most changes to the open-source projects of Simon Willison, an independent software developer, now arrive as pull requests that AI coding agents open. Willison found the agents' descriptions accurate but so long that readers, the author included, stop reading. Each one now leads with a short opening that Willison has checked, and the agent's full text sits in a collapsed section.

Key points
  • The opening paragraph is the shortest text that tells a reader what the pull request is for and whether it deserves more attention, sometimes with a screenshot.
  • The agent's full description follows inside a collapsible HTML section, which GitHub supports, labelled with the AI model's name and often its effort setting.
  • Willison does not always write the opening, but always chooses it and vouches that it is worth the reader's time.
Implication

An engineering team that receives pull requests from AI agents could ask for a short opening checked by a person. Reviewers then know which part carries human judgement. This is an inference from one developer's practice on personal projects.

Plays
Engineering

Open each AI agent's pull request with a short summary a person has checked, and fold the agent's full description into a collapsed section.

Derived by Working Surface from the article; more in the Playbook. Source line: My promise to the reader is that the short opener will represent my human judgement as to what is definitely worth their time.

Source issue

10 November 2026: as AI agents do more of the building, teams are naming the one point where a person's judgement must sit: approving a merge or release, vouching for a summary or defining good output.