Let Claude Code steer a Chrome extension for a week
They put Claude Code in charge of a free Chrome extension for nightly research, coding, and reporting, while keeping a human check before anything public went out.
- What they did
- The developer gave Claude Code a hand-edited markdown playbook and had it operate from the desktop app on a nightly schedule. With Playwright and a logged-in Chrome profile, it looked at the Chrome Web Store dashboard, read comments, checked subreddits, outlined posts, updated the extension, and saved a pinned Notion report. If a post or other public change was ready, it still had to stop and wait for the human to answer “go.”
- How it went
- In the first week, the agent spotted a near-duplicate featured extension, prompted a rename, cleaned up stale privacy copy, and built a new release with onboarding and a window-stashing feature. After 67 unit tests and 55 end-to-end tests, that release went to store review. The project numbers stayed very small: 4 weekly users, 14 installs over 30 days, 0 reviews, $0 spent, and one release sat for five days when Google demanded another sign-in.
- Tips
- If the account depends on Google, expect login prompts to interrupt the flow every few days. It helps to surface failed logins right away in the nightly report so they are not buried.
Try it yourself with Claude Code
Take the role of CEO for my [project name], a free [Chrome extension/app/tool], for the next 7 days. Using the codebase in [folder or repo], the live dashboard, and any recent feedback from [comments/forums/community spaces], decide the three most important tasks for tomorrow, explain why they matter, and begin work on the first one. Before you publish, submit, rename, delete, or send anything, show me the exact draft or change and wait until I reply “go.” End with a short status note plus any blockers, especially login or permission issues.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Make task handoffs require a commit ID
A developer paired a YAML todo script with a Claude Code helper that fetched the next job and closed it only after a commit ID was provided.
Match a pull request to repo style
A developer used Claude Code to bring a pull request in line with a repo’s style, but it only altered comments and left the code unchanged.
Use Claude Code for a carefully checked migration
A developer used Claude Code to update a reset feature, but the review work and cleanup erased most of the speed gains.
Share files between sandboxed Claude Code agents through MCP
A developer connected sandboxed Claude Code instances to a virtual file system so they could pass files through MCP without direct access to AWS.
Use Claude Code with Ollama for local coding help
They configured Claude Code to talk to a local Ollama model and used it on a small Java project inside IntelliJ.
Test OpenCode and DeepSeek for cheaper coding
A developer replaced Claude Code with OpenCode plus DeepSeek and said the experience felt similar for much less money.