Kontour Surface maps what a product claims, what proves it, where trust is missing or stale, and what humans or AI agents need to verify next.
It is not a promise of perfect truth. It is infrastructure for evidence-backed systems: claims, evidence, verification policies, freshness, conflict state, and inspectable reports.
AI makes plausible output cheap. Products now need a durable way to answer:
- What are we claiming?
- What evidence backs that claim?
- How was it verified?
- How long is that verification valid?
- What changed or conflicted since then?
- What should a human or agent prove before relying on it?
veritas: developer proof for repo surfaces, policy packs, proof lanes, and evidence artifacts.campfit: public-data verification through crawl evidence, field attestations, review flags, and freshness.taxes: high-stakes fact verification through extraction, resolution, verified facts, citations, and review signals.
These are proof customers for the trust layer, not the whole Kontour AI business.
npm install
npm run verify
npm run surface:report
npm run docs:buildThe first prototype reads examples/surface-fixtures.json, derives claim statuses, and emits a local trust report.
Veritas evidence can be imported directly:
npm run build
node bin/surface.mjs report --adapter veritas --input examples/veritas-evidence.json --format summaryCampfit and taxes trust exports use the same report contract:
node bin/surface.mjs report --adapter campfit --format summary
node bin/surface.mjs report --adapter taxes --format summarysrc/: TypeScript trust kernel and CLI helpers.src/adapters/: importers that map product-specific proof artifacts into the Surface kernel.schemas/: JSON schema contracts for Surface records.examples/: validation fixtures fromveritas,campfit, andtaxes.tests/: unit and fixture coverage.docs/: narrative docs used by the static Pages build.scripts/build-pages-site.mjs: dependency-free GitHub Pages builder.
This repo is intentionally local-first. Hosted dashboards, accounts, storage, and full production adapters are later stages. The first milestone is proving that different product types can share one inspectable trust report without losing evidence detail.