feat: publish ProveKit v1 benchmark story

[dcbuild3r]

Summary

• publish the frozen ProveKit v1 benchmark campaign results across mobile, browser, payload, proof-size, and low-memory scenarios
• make unsupported and out-of-memory results explicit so comparisons do not imply measurements that were not completed
• update the live benchmark demo to a source-pinned ProveKit v1 browser runtime and regenerated SHA-256 proving artifacts
• record runtime provenance and align the demo's Noir dependencies with the v1 toolchain

Validation

• astro check (0 errors; 1 pre-existing hint)
• Vitest (3/3 tests passed)
• production Astro build
• browser proof generation and local verification (VERIFIED; 16,429 constraints; 31,446 witnesses; ~797.3 KB proof)

Preview acceptance

The Vercel preview will be checked after deployment, including the benchmark page and hosted live proof demo.

[dcbuild3r]

Vercel preview: https://provekit-q516t0uhy-world-foundation.vercel.app/benchmarks

Hosted acceptance passed: the page returned HTTP 200 and the live SHA-256 demo completed with VERIFIED (16,429 constraints, 31,446 witnesses, 798.9 KB proof in the verification run).

[dcbuild3r]

Axis clipping fixed in 7df6cd9. Updated Vercel preview: https://provekit-c1s5gwa1y-world-foundation.vercel.app/benchmarks

Hosted browser verification now renders the complete monotonic payload axis: 0, 450, 900, 1350, 1800 MB.

[dcbuild3r]

Summary refactored in 896c159 to compare Passport P1 across six categories in fixed narrative order: iPhone, Moto E15, browser, payload, proof size, and Moto E15 peak RSS.

Updated Vercel preview: https://provekit-700ab8nv6-world-foundation.vercel.app/benchmarks

Hosted browser verification confirms the six rows render in that order.

[dcbuild3r]

Raw Results now points to the canonical benchmark branch in b36d515: https://github.com/worldfnd/provekit/tree/v1_benchmarks/benchmarks/v1

Updated Vercel preview: https://provekit-js8ue8d81-world-foundation.vercel.app/benchmarks

Hosted browser verification confirms the button resolves to the canonical URL.

[dcbuild3r]

Updated the landing page with the publication benchmark story and deployed the latest PR head.

Preview:

• Landing: https://provekit-gzovzhmre-world-foundation.vercel.app/
• Benchmarks: https://provekit-gzovzhmre-world-foundation.vercel.app/benchmarks

Acceptance:

• landing exposes the canonical 2–3s iPhone, <30s Moto E15, <3 MB payload, <500 MB peak phone RSS, and <1 MB proof claims
• timing copy explicitly includes witness generation through serialized proof
• benchmark page has one ProveKit v1 live demo and links the canonical raw results
• hosted browser proof completed and verified locally (VERIFIED)
• bun run check, bun run test, and bun run build pass

[dcbuild3r]

Updated the iPhone timing chart treatment in f265fbc:

• y-axis now focuses on 0–16s so the meaningful differences remain legible
• Circom WebAuthn keeps its true 151.42s value but renders as an intentionally clipped bar with a double diagonal break
• the overflow treatment is generic and only enabled for this chart

Updated preview:

• Landing: https://provekit-3d6t5xtbu-world-foundation.vercel.app/
• Benchmarks: https://provekit-3d6t5xtbu-world-foundation.vercel.app/benchmarks

[dcbuild3r]

Made missing benchmark runs explicit in 86c4e81: OOM now stays visible without hover while numeric result labels remain hover-only.

Updated preview:

• Landing: https://provekit-oawoocujz-world-foundation.vercel.app/
• Benchmarks: https://provekit-oawoocujz-world-foundation.vercel.app/benchmarks

Hosted browser verification confirmed the Moto E15 Circom WebAuthn OOM label has computed opacity 1.