Skip to content

feat: add an agent results page that reads the hosted benchmark data - #32

Merged
Teakowa merged 1 commit into
mainfrom
feat/agent-results-page
Oct 5, 2026
Merged

Teakowa merged 1 commit into
mainfrom
feat/agent-results-page

Conversation

@e54-bot

@e54-bot e54-bot commented Oct 4, 2026

Copy link
Copy Markdown
Contributor

Fixes #31

Summary

  • /results and /zh-CN/results fetch https://releases.wrightkit.dev/bench/latest.json in the browser and show a ranked table: per-track score bar with 95% interval, agent program/version, model, effort, "Tied with top" and "Provisional" marks.
  • Plain-language reading guide, "what was run" block (Wright and suite versions), empty and older-format states.
  • Header gets a Results link; nav and language switcher work from the results page. Home <head> is extracted into PageHead and reused; sitemap and verify-build cover the new pages.
  • Adds vitest (pnpm test, wired into CI).

Open points

  • wright-agent-results/v1 is not committed in wrightkit/wright yet (publish issue is open). src/lib/bench.ts reads: schema, environment.{wright,suite}, agents[].{agent.{program,version},model,effort,tracks.{workshop,opy}.{score,ci95,provisional}}. Field names follow the harness score card and must be reconciled with the committed schema.
  • The bucket must allow cross-origin GET from wrightkit.dev.
  • Rows are ordered by Workshop then OverPy score (never averaged); ties are measured against the first row per track, as the issue states.

Test plan

  • pnpm test: fixture renders scores, intervals, tie marks in en and zh-CN; empty/unsupported states
  • pnpm check, pnpm build (incl. verify-build)
  • 375px width / dark mode visual check (not run: headless browser could not reach the local server in my environment)

Adds /results and /zh-CN/results. The browser fetches
releases.wrightkit.dev/bench/latest.json (wright-agent-results/v1) and
shows a ranked table with per-track score bars, 95% intervals, and tie
marks, plus empty and older-format states.

Fixes #31
@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying with  Cloudflare Workers  Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status Name Latest Commit Preview URL Updated (UTC)
✅ Deployment successful!
View logs
wrightkit-dev bb64f6e Commit Preview URL

Branch Preview URL
Oct 04 2026, 04:36 PM

@Teakowa
Teakowa merged commit 633ccd3 into main Oct 5, 2026
2 checks passed
@Teakowa
Teakowa deleted the feat/agent-results-page branch October 5, 2026 12:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Add an agent results page that reads the hosted benchmark data

2 participants