Public Site

Portable publication surface for the LockedIn FDE Benchmark v0.1 Public Draft and the separately labeled 0.2.0-proposal interface preview. The current release context is AI-native Forward Deployed Engineering; FDE remains the broader operating model, and later non-AI context packs require separate practice analysis and validation.

The static public-draft deployment is live at fdebenchmark.org. It does not publish an official score, credential, ranking, research finding, or operational intake channel, and is indexable on its canonical host. Indexing does not activate the separate official-assessment release gates.

The design direction is an editorial measurement instrument: warm paper, exacting rules, high-contrast evidence panels, and restrained signal color. It intentionally avoids leaderboard theater and unsupported score visualization.

The Phase 2 source preview adds:

a five-state, score-suppressed trust ladder with recency and public-projection boundaries;
a non-person profile anatomy rather than a fabricated practitioner;
12 separate capability, operating, and evidence-meta observation axes with no scalar result or proposed weights;
an unmistakably synthetic metro-participation fixture with candidate 30-subject and five-organization suppression, no country comparison, and no person lookup;
behavioral integrity routing that is expressly geography- and prestige-neutral; and
a human review, correction, and appeal flow with no operational intake.

These surfaces demonstrate product structure only. They do not amend v0.1, authorize a service, activate a register, or establish empirical findings.

Product routes

/ — benchmark overview and research highlights
/explorer — searchable 10-domain/55-family public ontology and proposed evidence design
/assessment — synthetic evaluator workspace
/research — proposed multi-cohort field-research program
/publications — method records, benchmark precedents, field notes, and market intelligence
/method — public benchmark method paper
/authority — publication identity, evidence state, accountability, and institutional-firewall records
/pilot — proposed applied-assessment laboratory with activation kept fail-closed

Legacy ?view= URLs remain supported for existing links.

Commands

npm install
npm run dev
npm run typecheck
npm test
npm run build
npm run check

The in-site document reader bundles and sanitizes the canonical Markdown sources from the fde-benchmark repository, including the Phase 2 proposal contracts and research agenda. Stable ?doc=<key> links keep public methodology, governance, status, chart, proposal, and publication-template references in the reader instead of navigating to raw build assets.

The domain interface imports the versioned v0.1 ontology JSON directly. Publication tests fail if the rendered domain/family total drifts from the 10-domain, 55-family source artifacts or if source-status, disclosure, document-link, and baseline accessibility requirements regress.

npm run check runs source-claim checks, a rendered Chromium suite, axe checks against WCAG A/AA tags through 2.2, keyboard and tab semantics, mobile and 200%-equivalent reflow checks, strict TypeScript, and the production build. Automated accessibility checks complement rather than replace manual assistive-technology and participant testing required before publication.