Repository navigation
chore(brand): rename Browser Eval to Webstone - #273
lanyue-llk wants to merge 1 commit into
Conversation
Rename the project brand to Webstone, "the touchstone for your agent's web tools", across docs, dataset cards, report labels and GitHub links. Add the name's idea and slogan to both READMEs and replace the hero image. Machine identifiers stay: the browser-harness-eval package and CLI, BROWSER_EVAL_* variables, ~/.browser-eval/, schema IDs and the hash-bound Web Fetch scorer source, so saved runs and published reports keep working.
![lexmount-ocr[bot]](https://avatars.githubusercontent.com/u/277637559?s=60&v=4){kind=small}
There was a problem hiding this comment.
This review found 1 issue(s), including 1 blocking issue(s).
| labels = page.select('[data-browser-eval-release]') | ||
| if (len(labels) != 1 or labels[0].get("data-browser-eval-release") != identity["version"] | ||
| or labels[0].get_text(" ", strip=True) != "Browser Eval v" + identity["version"] | ||
| or labels[0].get_text(" ", strip=True) != "Webstone v" + identity["version"] |
There was a problem hiding this comment.
P2: This rejects intact reports generated before the rename: their bound homepage label is Browser Eval v<version>, so check_build_version now raises even when the manifest, identity file, version and link are all valid. Both check_report_identity and Web Fetch's check_release invoke this check on saved reports. This breaks historical-report verification, contrary to AGENTS.md's requirement to preserve published reports and keep frozen evidence verifying. Keep emitting Webstone for new builds, but accept the legacy brand label during validation as well, retaining the exact version and link checks. Add a regression test that validates an unchanged pre-rename report.
Renames the project from Browser Eval to Webstone: The touchstone for your agent's web tools.
Assayers tested gold on a touchstone: rub the metal on the dark stone and the streak shows what it is. Webstone does the same for the browsers and web tools that AI agents rely on. Every tool runs the same frozen tasks, every call is kept as evidence, and the report shows what each one left behind.
Changes
lexmount/webstone. GitHub redirects the old URLs after the repository rename.Deliberately unchanged
Machine identifiers that saved data, hashes or published packages depend on: the
browser-harness-evalpackage and CLI,BROWSER_EVAL_*variables,~/.browser-eval/,data-browser-eval-*attributes, schema IDs, HTTP headers, Docker labels, andsrc/lexbench_webfetch/, whose source hash is bound to released scorer versions. Recorded data underresearch/and*/data/*.jsonis untouched.Checks
make docs-check lint-checkpass. Full Python suite: 22840 passed; the 12 local failures (Node Markdown converter, CI-gate, Playwright tool) also fail on clean main and depend on the local environment.