Null Lens — Enterprise API Docs

curl https://null-core.ai/api/lens-public-key

Example response:

{
  "public_key_id": "lens_2026_v1",
  "algorithm": "ed25519",
  "public_key": "BASE64_PUBLIC_KEY"
}

The verifiable payload is the canonical AIR object only, in this exact field order: record_id, timestamp, version, scope, boundary.

{
  "record_id": "air_36b4a07e-d239-46d3-8ce9-d285a83f71e4",
  "timestamp": "2026-03-11T22:15:03.759Z",
  "version": "v1.0",
  "scope": "internal audit posture",
  "boundary": "Strictly limited to the expressed scope. No additional domains inferred."
}

{
  "signature": "LRLoXyk5lSZ/jqtqKov1iRyWbtsVim01nRHX8DZkGJbBMeIxq25rL+X3EB9gg0hrhne65tNPSEJAx7ge7CgpDA==",
  "public_key_id": "lens_2026_v1",
  "algorithm": "ed25519"
}

Install the verifier dependency first: npm install tweetnacl

import nacl from "tweetnacl"
import crypto from "crypto"
import { Buffer } from "buffer"

const air = {
  record_id: "air_36b4a07e-d239-46d3-8ce9-d285a83f71e4",
  timestamp: "2026-03-11T22:15:03.759Z",
  version: "v1.0",
  scope: "internal audit posture",
  boundary: "Strictly limited to the expressed scope. No additional domains inferred."
}

const signature =
  "LRLoXyk5lSZ/jqtqKov1iRyWbtsVim01nRHX8DZkGJbBMeIxq25rL+X3EB9gg0hrhne65tNPSEJAx7ge7CgpDA=="

const publicKeyBase64 = "BASE64_PUBLIC_KEY"

const canonical = JSON.stringify({
  record_id: air.record_id,
  timestamp: air.timestamp,
  version: air.version,
  scope: air.scope,
  boundary: air.boundary,
})

const hash = crypto.createHash("sha256").update(canonical).digest()
const sig = Buffer.from(signature, "base64")
const pubKey = Buffer.from(publicKeyBase64, "base64")

const valid = nacl.sign.detached.verify(hash, sig, pubKey)

console.log("Signature valid:", valid)

Signature valid: true

When Lens cannot safely resolve an explicit scope due to unsafe, malformed, adversarial, or non-resolvable input, it may return a containment AIR that preserves the response contract while narrowing scope to a safe fallback.

{
  "air": {
    "record_id": "air_xxx",
    "timestamp": "2026-03-11T22:15:03.759Z",
    "version": "v1.0",
    "scope": "Contained Input",
    "boundary": "Strictly limited to the expressed scope. No additional domains inferred."
  },
  "integrity": {
    "payload_hash": "sha256:...",
    "signature": "...",
    "public_key_id": "lens_2026_v1",
    "algorithm": "ed25519"
  }
}

This indicates deliberate containment, not silent corruption. Lens preserves the signed response model while preventing unsafe, adversarial, or non-resolvable input from influencing downstream systems.

Downstream systems should treat a containment AIR as non-actionable for downstream execution or enforcement. Do not use scope: "Contained Input" as authorization to proceed.

Halt execution and route the request for human review or retry. Repeated containment on the same or similar input is a signal of adversarial input, prompt injection, or transport corruption rather than a transient inference failure.

• Halt downstream execution — containment AIRs are not authorization to proceed.

• Inspect the request for prompt injection, override attempts, or malformed transport payloads.

• Retry once if the containment response is unexpected and the input appears legitimate.

• Verify the returned signature before relying on the response downstream.

• Route to human review if containment persists. Repeated containment on similar inputs indicates adversarial activity or transport corruption — not a transient inference failure.

Lens is designed for high-availability enterprise deployments. Use standard retry logic for transient network or upstream failures, and treat verification as the final acceptance check before downstream use.