Current engineering proof

What Site Machine can prove today

The public wedge is not a vibe claim. These are the current operator-backed proof points for Sara, launch traction, vertical templates, selective iteration, and deployment reliability on staging.

Live facts, roadmap gates, and operator commands are separated before the fold.

Live dataRoadmap gateOperator command

Sara learning loop

Live data
5

multilingual Sara regression briefs passed

Live Sara proof: multilingual regression briefs passed across 5 regression briefs max, with 1 deployed editor-origin prompt(s) and archived per-brief score evidence. Current deep proof bar is five Persian, Arabic, Hindi, and Gujarati briefs.
Operator commandnpm run sara:proof:staging:editor:deep:required

Persian/Arabic launch traction

Roadmap gate
0/100

launch-sourced brief submissions

The Persian and Arabic launch pages now preserve acquisition source into signup audit events, direct job payload attribution, and pre-auth launch brief lead attribution. Current proof: 0/100 launch-sourced brief submissions; 0/30 completed launch-sourced sites. Campaign proof: x1-persian-launch 0, x1-arabic-launch 0, Not recorded 0.Next action: Convert 100 more launch-sourced briefs and 30 more completed launch sites.
Operator commandnpm run x1:launch-proof:staging

Selective iteration

Live data
unknown

live selective iteration token ratio

Live 30-day proof: 0 selective iteration job(s), but the token ratio is unknown because parent usage is missing. Selective iteration cohort is too small: 0/10 required jobs.
Operator commandnpm run iteration:proof:staging

Editor-agent loop

Roadmap gate
1/3

active editors with merged agent proposals

The multi-editor mechanics are shipped: proposal authorship, bounded concurrent processing, PR attribution, and the leaderboard. Current proof: 1/3 active editors have at least 1 merged editor-agent proposal.Next action: Onboard 2 more active editors with merged editor-agent PRs.
Operator commandnpm run x4:editor-loop-proof:staging

Founding-editor queue

Roadmap gate
0/3

selected pilot candidates

The public /founding-editor form now feeds the N2 proof path. Selected, sent, and first merged editor-agent PR evidence count toward the active roadmap proof without exposing applicant identity. Current application proof: 0/3 selected, 0/3 sent, first PR not merged. Missing criteria: Persian-speaking or RTL reviewer, design background, shipped websites for paying clients. Application sources: X4 founding-editor outreach 0, Public founding-editor page 0.
Operator commandnpm run roadmap:proof:active:staging:required

Sara-origin proposals

Roadmap gate
100%

merged proposals from quality signals

The quality-signal bridge can queue Sara's draft as a normal editor proposal, run one editor-agent cycle, and open the resulting GitHub PR with npm run quality:proposal-cycle:open-pr:staging -- --first-operator. Measure closure with npm run l2:agent-origin-proof:staging. Current proof: 1/1 merged proposals originated from Sara quality signals (100%).Next action: Keep Sara-origin proposals at or above 50% of merged editor proposals.
Operator commandnpm run quality:proposal-cycle:open-pr:staging -- --first-operator

Vertical template depth

Operator command
12

production template previews guarded

12 production-capable vertical template previews are live from the registered template set: restaurant menu, photographer portfolio, quant strategy, single-product dtc, freelancer case studies, real estate listings, personal brand writer, wedding and event, clinic and dentist, bilingual local services, construction project management, and contractor services.
Operator commandnpm run staging:smoke

L4 template expansion

Roadmap gate
12/10

production vertical templates registered

The L4 bar is ten registered specialist templates before Site Machine can claim durable depth beyond the first wedge set. Current proof: 12/10 production vertical templates are registered.
Operator commandnpm run l4:template-depth-proof

L4 build depth

Operator command
12/12

registered templates build-backed

12/12 registered production vertical templates are build-backed.
Operator commandnpm run l4:template-build-depth-proof

L4 specialist builds

Live data
0/7

specialist templates completed on staging

Live 30-day proof: 0/7 specialist templates have completed build artifacts; missing real-estate-listings, personal-brand-writer, wedding-event-landing, clinic-dentist, local-services-bilingual, construction-project-management, and contractor-services. This proof is not counted as X2 prompt-provenance; prompt provenance is measured by the specialist prompt-backed card. real-estate-listings has 0/1, personal-brand-writer has 0/1, wedding-event-landing has 0/1, clinic-dentist has 0/1, local-services-bilingual has 0/1, construction-project-management has 0/1, contractor-services has 0/1.
Operator commandnpm run l4:specialist-build-run-proof:staging

L4 specialist prompts

Live data
0/7

specialist templates prompt-backed on staging

Live 30-day proof: 0/7 specialist templates have prompt-backed jobs; missing real-estate-listings, personal-brand-writer, wedding-event-landing, clinic-dentist, local-services-bilingual, construction-project-management, and contractor-services. real-estate-listings has 0/1, personal-brand-writer has 0/1, wedding-event-landing has 0/1, clinic-dentist has 0/1, local-services-bilingual has 0/1, construction-project-management has 0/1, contractor-services has 0/1, real-estate-listings has 0/1 prompt-backed, personal-brand-writer has 0/1 prompt-backed, wedding-event-landing has 0/1 prompt-backed, clinic-dentist has 0/1 prompt-backed, local-services-bilingual has 0/1 prompt-backed, construction-project-management has 0/1 prompt-backed, contractor-services has 0/1 prompt-backed.
Operator commandnpm run l4:specialist-prompt-proof:staging

Release reliability

Operator command
operator

operator-verified staging guardrail

Staging streak proof is operator-verified because the public runtime cannot call the GitHub CLI; run the command below for the live GitHub Actions streak, URL smoke, and build-pipeline smoke.
Operator commandnpm run roadmap:proof:active:staging:required

N1, N3, N5, X2, X3, and X5 are closed with N2/N4 parked.

The current proof board passes with the parked gates excluded. That means the learning mechanism, doc drift fix, creator quality surface, vertical-template prompt proof, selective iteration fast path, and staging reliability proof are closed. Future proof bars are measured publicly. Open future bars: X1 and X4. Closed future bars: L2 and L4. The remaining parked work is business/operator configuration and external proof rather than hidden engineering uncertainty; N2 selected/sent application counts are now visible above.

X1 and X4 stay measurable as open future bars.

The future proof report aggregates launch traction, multi-editor loop volume, Sara-origin proposal share, and ten-template depth without weakening the active deployment gate. Closed future bars: L2 and L4. Run npm run roadmap:proof:future:staging to see the staging DB-backed snapshot, run npm run roadmap:proof:future:production to verify the live production board on sitemachine.ai, or inspect the public JSON at /api/proof/future. The public API contract is available at /api/openapi.

N2
Founding-editor external proof

The mechanism exists, and the public pilot page is now shareable at /founding-editor, but the public proof still needs one real external editor to merge one product-improving PR through the agent loop.

N4
Annual Stripe checkout proof

Annual checkout code is wired. The remaining proof is production Stripe yearly price configuration and one real annual checkout. Verify production readiness with npm run stripe:annual-proof:production.

The wedge is visible. Now it needs volume.

The next proof step is real editor and creator repetition: more regression briefs, more real examples per vertical, and one external editor shipping through the loop.