What it's for
If you publish a long-running podcast, a video lecture series, or a book-length text, readers often want to find where in the corpus a topic is discussed. Index Engine builds that index — alphabetized, timecoded, searchable, gated.
Adding an entry manually
- Go to Wigmorean → Index Engine → Entries
- Click New Entry
- Enter the keyword (e.g., "consideration (contract law)")
- Add one or more references — each is a source, optional timecode, and a snippet
- Save
The auto-builder (Claude API)
Manually indexing a 200-episode podcast is untenable. The auto-builder does it:
- Give it a source — a post, a CPT archive, a transcript, or a video with captions
- Give it an indexing brief — "all legal terms of art", "every person mentioned", "every cited case"
- It calls the Claude API, extracts candidates, and proposes entries with timecodes
- You review and approve — accept, edit, or reject each candidate
- Approved entries land in the public index
The brief is reusable — save it as a preset so future episodes index themselves against the same rules.
Drawer UI
On the frontend, the index opens as a drawer sliding in from the right. The drawer has three panels:
- A–Z rail — jump to any letter
- Entry list — every keyword starting with the selected letter
- Reference panel — clicking a keyword shows all its references, each with a deep-link to the source at the right timecode
Drop the drawer into any page with [wig_index].
Timecodes
When the source is a video or audio file with captions, timecodes are extracted automatically. The deep-link opens the media at that exact moment (via the Media Connector if you're using one).
Access gating
Index entries can be gated by:
- User role (subscribers only, members only)
- Product purchase (has bought course X)
- Active subscription (paid tier)
Non-entitled visitors see the keyword but not the references — a teaser that converts well to paid upgrades.