Turn a multi-year meeting archive into text files for a notebook tool
Idea + starter prompt, not run by us yet
For: Researchers and community members who need an archive turned into searchable text
Jeff opened a public minutes archive and had Claude in Chrome (the Haiku model) visit each 2022 to 2024 meeting page and copy the text as markdown. He tried one meeting first, dropped to a month at a time when a whole year was too much, and spot-checked the results. Three yearly files then went into NotebookLM, after 3 to 4 hours across three sessions.
Steps
- Open the archive page and test the prompt on a single meeting first.
- Ask Claude to copy the text exactly, as markdown inside a code block, with no summarising.
- Work one month at a time. The agent stalled after three to four months of a year.
- Add each month's text to its year's file, then spot-check some entries against the original pages.
- Plan for usage limits. One limit forced a four-hour wait.
Starter prompt
Trial first: copy only one meeting, the first one for [month and year], into a file called [trial-minutes.md]. Stop there and wait for me to check it against the page. I have the public meeting minutes archive open in this tab: [URL]. For the meetings in [month and year], visit each meeting page and copy the minutes text exactly as written, as markdown inside a code block. Do not summarise or fix anything. Put the meeting date and page title above each block. This job only reads public pages and writes to [output folder]. Stop and wait for my yes before moving on to the next month. Do not log in to anything, and tell me if a page fails to load rather than skipping it silently.
Original post: lists.w3.org · found via agentusecases.ai. Written up in our own words. We haven't run it, so test it on made-up data first.