How to Maintain AI-Generated Chrome Extensions When Web UIs Update (Part 4)
This article was originally published on e-shikumi-labo . Hello, I'm Shin from e-Shikumi-Labo. This is the final installment (Part 4) of "Systematized Thinking," where we use AI to build our own tools and independently maintain them. So far, we have discussed creating a prototype that automatically saves Gemini chat logs, converting them to Markdown for Obsidian integration, and elevating it to a…
This is the concluding part of a series on building and maintaining AI-generated Chrome extensions, specifically focusing on the challenges and countermeasures that arise from web UI updates. The key points discussed include:
1. Data extraction limitations: Extracting data from web pages is only possible if that data is visible in the browser's screen or DOM. Forcing extraction of data not present in the screen leads to complex guesswork and potential issues.
2. Anticipating UI changes: Websites and platforms frequently update their design. When your tool relies on extracting data from these sites, it might suddenly stop working due to these changes. This is not a tool defect, but a consequence of relying on an external platform.
3. Designing for separation and repair: The Chat Saver for Gemini tool separates the DOM parsing process into an independent file called parser.js. This allows for easier fixes when UI changes occur, as only the designated parts of parser.js need modification, rather than rewriting the entire code.
4. Seeking AI assistance for fixes: When UI updates break your tool, you don't need to directly modify the code yourself. Instead, you can present the current parser.js code and the new HTML elements to an AI assistant, asking it to correct the selector parts. The AI can quickly generate the necessary modifications.
5. Building self-reliance in AI tooling: The ultimate goal is to develop a system where humans understand the overall structure of the tool (input, process, output) and can anticipate potential issues. By recognizing dangerous code generated by AI and providing precise instructions to correct it, humans can maintain the tool even when facing screen updates or other changes.
This self-reliance allows individuals to build and control AI systems tailored to their specific needs, rather than solely relying on the tools themselves.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
