Skip to content

Home

Jump to bottom
tavlean edited this page Jul 2, 2026 · 3 revisions

RankedAGI extension wiki: documentation, roadmap, and status live HERE — the extension repo stays doc-free (it gets PR'd to the Raycast Store verbatim, same rule as DevServers).

Each page covers one thing. Load the page you need; skip the rest.

The one-paragraph picture

The extension puts rankedagi.com in the launcher: Search Models (ranked list, full score detail), Search Benchmarks (per-benchmark model rankings), and four AI tools (@rankedagi in Raycast AI). Data flows one way: Tav edits data/*.json through the site's dev-only admin → every write passes through the site's recompute choke point → the next Cloudflare Pages deploy prerenders /api/export as a static JSON → the extension fetches it (cached, stale-while-revalidate) and does all search/rank/format client-side. No backend, no keys, nothing to operate.

Pages

  • Roadmap: the single source of truth for plan + status — phases R1–R4, upcoming-site-features track, decisions locked with Tav.
  • Data Contract: /api/export — exact shape, the additive-only evolution rule, the empty-levels rule, how the extension consumes it, local-dev testing. Site-side counterpart: docs/api-export.md in the site repo.
  • Architecture: extension file structure, the two data paths (React hook vs AI-tool loader), formatting rules that bite (percent fractions, benchmark names), store conventions.
  • Worklog: session history — what changed, why, gotchas.
  • Raycast Docs Research: the cited 2026-07-02 sweep of official Raycast docs that the architecture follows.

Repos

  • Extension: ~/Development/RaycastExtensions/rankedagi (npm package at root; ray develop / npm run build / npm run lint)
  • This wiki: ~/Development/RaycastExtensions/rankedagi.wiki — local git repo for now; when the GitHub repo for the extension exists, create its wiki and push this repo there (same flow as DevServers.wiki, remote github.com/tavlean/<repo>.wiki.git)
  • Site (data source): ~/Development/Tavlean/RankedAGI — see its docs/api-export.md and .claude/PROJECT_BRIEF.md; its own roadmap (score model v2) is independent and runs in parallel sessions

Clone this wiki locally