Agents In UseThese software & devops AI agent use cases include accounts involving Claude, Codex, Cursor. Each entry links its source and describes the task, reported outcome and human checkpoint.
Examples include “Claude Code builds a local video compression interface” (Claude); “Claude Code builds and browser-tests a GIF optimizer” (Claude); “Claude Code ports an inpainting model into a browser” (Claude).
In “Claude Code builds a local video compression interface”, Willison chose the recording and evaluated output options. Publication approval boundaries are not described. In “Claude Code builds and browser-tests a GIF optimizer”, Willison watched progress and added requirements for reproducible build files and upstream credit.
13 use cases
simonwillison.netWillison wanted to optimize a phone recording before publishing it. He used Claude Code for web to build a browser tool around FFmpeg compiled to WebAssembly, with multiple output presets.
Added
Claudevia Claude Codesimonwillison.netWillison asked Claude Code for web to compile Gifsicle to WebAssembly and build a preview interface. The agent used Rodney to exercise the page and fix UI bugs during implementation.
AddedClaudevia Claude Codesimonwillison.netWillison delegated a Moebius model port to Claude Code using local repositories and a research note. He tested the browser UI and returned errors while the agent iterated on ONNX and WebGPU support.
AddedClaudevia Claude Code
gist.github.comWillison published the Codex session used to build the Datasette project blog. The agent inspected the existing site, moved conflicting database routes and implemented blog behavior with local checks.
Added
Codexsimonwillison.netSimon Willison gave Codex a specific defect in his Django blog: social preview descriptions exposed raw Markdown. Codex changed templates and added a passing test in a configured PostgreSQL environment.
AddedCodexsimonwillison.netWillison supplied an existing parser and an independent compatibility suite, reviewed an API plan, then let Codex implement a JavaScript port through repeated tests and commits.
AddedCodex
kentcdodds.comDodds used Cursor with GPT-5.4 to add lexical retrieval to his semantic search. Production blocking exposed an architectural flaw, so later agent work moved search into a unified Worker.
Added
Cursorkentcdodds.comDodds delegated an audio pipeline migration to Cursor. He rejected an initial design that preserved a separate service and used another agent run to make the sandbox part of the existing queue Worker.
AddedCursorsimonwillison.netWillison recorded a real build with Claude Code for web, turning pasted terminal output into shareable HTML. The companion article explains the repository references and the workflow behind the recording.
AddedClaudevia Claude Codesimonwillison.netWillison recorded upgrading datasette-checkbox for a changed permissions API. He first repaired a test that missed a real page failure, then gave Codex the failing runner and upgrade documentation.
AddedCodexAn illustrative video entry showing how an agent workflow could be reviewed, with a deliberate human checkpoint.
Added
OpenClawInspect proposed dependency updates, review upstream changes, and test compatibility in a disposable environment.
AddedOpenClawInspect a repository, reproduce a reported bug, and prepare a tested patch for review.
AddedOpenClawExamples include “Claude Code builds a local video compression interface” (Claude); “Claude Code builds and browser-tests a GIF optimizer” (Claude); “Claude Code ports an inpainting model into a browser” (Claude). Open each entry for the steps and reported outcome.
In “Claude Code builds a local video compression interface”, Willison chose the recording and evaluated output options. Publication approval boundaries are not described. In “Claude Code builds and browser-tests a GIF optimizer”, Willison watched progress and added requirements for reproducible build files and upstream credit.
The sourced entries in this category name Claude, Codex, Cursor. Open each account to see the tools used and the author’s reported result.
These accounts document particular runs rather than establish safety for every task. The recorded human checkpoints are: In “Claude Code builds a local video compression interface”, Willison chose the recording and evaluated output options. Publication approval boundaries are not described. In “Claude Code builds and browser-tests a GIF optimizer”, Willison watched progress and added requirements for reproducible build files and upstream credit. Review the linked source and permissions before trying an adapted prompt.