agency-docs-updater

Meeting documentation is automated by running a script that merges Fathom transcripts, YouTube embeds, and fact-checked summaries into a publish-ready MDX file. The script auto-detects lab numbers from filenames and consolidates all content into a single document.

Embedding YouTube links and transcripts into an MDX file is handled by passing the transcript file and YouTube URL to the update script. The script embeds video content and merges fact-checked summaries directly into the final MDX output.

Auto-detecting lab numbers from filenames is a built-in feature of the documentation update script. It reads the filename to determine the lab number and supports both auto-detection and explicit meeting numbering for generating the next meeting docs.

Before generating MDX meeting docs, you need a fact-checked summary file, a Fathom transcript file, and a YouTube URL. You generate the summary using claude-code-guide, then run the Python update script with these inputs to produce the document.

The meeting documentation workflow supports optional translation and version control through git. After editing the generated frontmatter, you git add, commit, and push the MDX file to trigger deployment pipelines and keep docs in sync with lab activity.

Frontmatter issues occur because the script enforces a frontmatter-driven entry that requires manual editing to finalize the title and description. You must generate the MDX file first, then edit the frontmatter to finalize metadata before committing and pushing.