Trust Infrastructure — Open source

Map what your
product claims.
Prove it.

AI increases the volume and polish of product claims. Kontour makes the evidence inspectable — what's verified, what's stale, and what still needs proof.

Explore Veritas → Explore Surface
verified proposed stale disputed unknown — every claim has a state
The Problem

AI agents are
confident.
Evidence is not.

Every AI agent asserts. Every feature claims. Every product says it does what it says. What's missing is the chain from assertion to evidence — and a way to see where that chain is broken, stale, or never existed.

"Evidence-backed confidence. Not certainty theater."

claim trace  ·  api.rateLimit
claim
Rate limit: 100 req/min per key
source: README, spec/rate-limit.test.ts
evidence
test output — 312 ms ago
proof lane: npm test
check
CI run #1847
6 days since last verification
policy
freshness window: 72h
⚠ window exceeded — requires re-verification
Two products. One trust layer.
v0.1.3
@kontourai/veritas

Veritas

BESPOKE LINT FOR AI AGENTS

Repo-local governance that tells AI agents exactly what changed violated your rules — before they finish work. Works with Claude Code, Cursor, Codex, and Copilot.

artifactsgovernance-blockif-changedproof-lanes
Explore Veritas →
v0.1.0
@kontourai/surface

Surface

THE TRUST MAP FOR AI-ERA PRODUCTS

Maps what your product claims, what evidence supports it, where trust is missing or stale, and what humans or AI agents need to verify next.

claimsevidencedriftfault-linescoverage
Explore Surface →
Philosophy
"Evidence-backed
confidence.
Not certainty theater."

Surface does not market itself as objective truth. It shows the evidence chain, the verification method, and the current trust state. If a claim is weak, stale, or disputed — the system makes that obvious.

Get started

Wire Veritas into
your repo in minutes.

Bootstrap your adapter, define proof requirements, and get lint-style feedback straight into your AI agent's context.

Read the docs → GitHub
veritas — setup

$ npm install -g @kontourai/veritas

added 1 package in 0.8s


$ veritas init

✓ detected repo shape

✓ wrote .veritas/repo.adapter.json

✓ wrote .veritas/policy-packs/default.policy-pack.json

✓ governance block added to CLAUDE.md, AGENTS.md


$ veritas shadow run

[feedback] 2 rules checked, 1 pass, 1 warn

WARN if-changed: src/api/ changed — tests/api/ must also appear

evidence written → .veritas/evidence/