agentsclimarketplace

K id age verification

Skill kidentify/skills/skills/k-id-age-verification

Implements k-ID age verification and assurance for all three CDK patterns: (1) standalone AgeKit+ with no session (single age-proof — UK OSA 18+, age-restricted downloads, Australia social-media minimum age), (2) unverified-adult verification after the gate in a sessioned integration, (3) per-permission threshold verification (Brazil ECA Digital loot-boxes / targeted ads, UK OSA 18+, Australia social media, any verifiedAgeThreshold regime). Covers /age-verification/perform-access-age-verification (with or without sessionId), /session/upgrade returning CHALLENGE_SESSION_UPGRADE_BY_AGE_ASSURANCE, iframe modal (never window.open), in-app prompt, new-tab return-path detection (server poll + BroadcastChannel + visibility wake), result via webhook or polling, race-free session update. References AgeKit+ waterfall and ConnectID. Use for AgeKit+, unverified sessions, verifiedAgeThreshold, highly effective age assurance, or "user verified but app never unlocked". Not the initial gate or consent (see sibling skills).From its SKILL.md

Install
npx -y skills add kidentify/skills --skill k-id-age-verification

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

17.7 KB, as published. Nobody here has run it

k-ID Age Verification and Age Assurance

This skill covers age assurance — proving a user's age (or that they're above a required age) to the standard a regulator requires. Three concrete patterns show up in k-ID integrations:

  1. Standalone AgeKit+ (no session). The product only needs to prove age once, for one decision, with no persistent k-ID session and no permission state. Typical for 18+ sites under UK OSA, age-restricted downloads, Australia social media minimum-age checks, or a single high-risk action. Call /age-verification/perform-access-age-verification without a sessionId, render the returned URL in an iframe, act on the result via webhook or GET /age-verification/get-status.
  2. Unverified-adult in a sessioned integration. /age-gate/check returns an adult session with ageVerificationStatus === 'unverified'. The session exists but privileged permissions are withheld until the user completes verification. Common for COPPA-style "prove you're not a minor" gates layered on top of a full k-ID integration.
  3. Per-permission threshold verification (in a sessioned integration). A specific permission carries a verifiedAgeThreshold (for example 18 for Brazil ECA Digital loot-boxes / targeted ads, 12 for Brazil direct marketing, 18 for UK OSA 18+ permissions, 16 for Australia social media). /session/upgrade returns a CHALLENGE_SESSION_UPGRADE_BY_AGE_ASSURANCE. The whole session isn't locked — just that one feature.

All three patterns use the same verification UX primitives (iframe modal, detection strategies, result via webhook or polling) and the same k-ID age-assurance methods (ID document, facial age estimation, AgeKey, credit card, email age estimation, regional providers such as ConnectID in Australia and Singpass in Singapore). The "highly effective age assurance" standard from UK OSA maps to the same endpoints — method selection happens server-side based on jurisdiction and Compliance Studio configuration, not in client code.

When to use

Use this skill when the user is:

  • Adding standalone AgeKit+ age assurance to an 18+ site, age-restricted download, or single high-risk action (Shape B of the router's two integration shapes).
  • Building the in-app verification prompt that asks a user to verify inside a sessioned integration.
  • Calling /age-verification/perform-access-age-verification or /age-verification/get-status in either shape.
  • Handling CHALLENGE_SESSION_UPGRADE_BY_AGE_ASSURANCE returned by /session/upgrade for a verifiedAgeThreshold permission.
  • Implementing polling, webhook handling, or BroadcastChannel detection for when the user completes verification.
  • Meeting "highly effective age assurance" requirements under the UK Online Safety Act, or equivalent thresholds under Brazil ECA Digital or Australia's social media age rules.
  • Diagnosing reports like "user verified but the app never unlocked the feature" or "webhook fired but our app never received it".

This skill is the whole integration for Pattern 1 (standalone AgeKit+). For Patterns 2 and 3 it composes with k-id-sessions-and-permissions.

Core steps

1. Detect which pattern applies

Pattern 1 — Standalone AgeKit+ (no session). The product has no k-ID session, no age gate, no permission-management UI. The trigger is a user-initiated action: they click a button on an 18+ site, attempt an age-restricted download, or tap "verify" on a social media sign-up flow. No state to check ahead of time — just call the verification endpoint when the user starts the flow. See AgeKit+ overview.

Pattern 2 — Unverified adult in a sessioned integration. /age-gate/check (or /session/get) returns:

{
  "session": {
    "sessionId": "...",
    "ageCategory": "adult",
    "ageVerificationStatus": "unverified",
    "permissions": [ /* minor-default permissions */ ]
  }
}

The session has real permission data (minor defaults, not emptiness). Treat the app as operational but with a persistent verification prompt and locked features until verification completes.

Pattern 3 — Threshold permission (sessioned). A specific permission in session.permissions carries a verifiedAgeThreshold:

{
  "name": "loot-boxes-paid-gameplay-impacting",
  "enabled": false,
  "managedBy": "PLAYER",
  "verifiedAgeThreshold": 18
}

The rest of the session is fine — only this feature is locked until the user proves they're at or above the threshold. Calling /session/upgrade with this permission returns CHALLENGE_SESSION_UPGRADE_BY_AGE_ASSURANCE (not the standard consent challenge).

All three patterns feed the same verification UX in steps 2–6. The trigger differs; the UX does not.

2. Call /age-verification/perform-access-age-verification correctly

Body shape matters. The age criterion is nested under criteria, never at the top level. sessionId is optional — include it for Patterns 2 and 3, omit it for Pattern 1.

Pattern 1 — Standalone AgeKit+:

POST /age-verification/perform-access-age-verification
Authorization: Bearer <server-only API key>
Content-Type: application/json

{
  "jurisdiction": "GB",
  "criteria": { "ageCategory": "ADULT" }
}

Add optional fields when useful:

  • subject.email to reuse a prior k-ID verification result if one exists for that email (avoids asking the user again).
  • subject.claimedAge if your product already collected a claimed age and wants to inform age estimation.
  • subject.id (temporary session ID or hashed user ID) to correlate multiple attempts for the same user across methods.
  • options.redirectUrl if verification will open in a browser or webview rather than an iframe — k-ID redirects there with verificationId and result query params on completion.

Patterns 2 and 3 — In a sessioned integration:

POST /age-verification/perform-access-age-verification
Authorization: Bearer <server-only API key>
Content-Type: application/json

{
  "sessionId": "...",
  "jurisdiction": "US",
  "criteria": { "ageCategory": "ADULT" }
}

Top-level age, ageCategory, or minimumAge are rejected with 400 in either shape. All age inputs live inside criteria.

Response always includes id, url, and shortUrl. Treat shortUrl as opaque; don't rebuild or re-shorten the full url. For full shape see docs.k-id.com/api/endpoints/perform-access-age-verification.

3. Show the verification in an iframe modal — never a new tab

<Modal width={480} heightClamp="min(720px, calc(100vh - 48px))">
  <Toolbar title="Verify your age" onClose={...} />
  <iframe src={verification.url} style={{ flex: 1, border: 'none' }} />
</Modal>

The iframe content is the UI. No padding around it, no ornamental header, no explanatory copy in your modal. Thin toolbar with title and close (x) only.

Never window.open(verification.url, '_blank'). Two reasons:

  1. On iOS Safari, popup windows are aggressively blocked and the verification never starts.
  2. Even when the popup works, detection becomes a polling nightmare (see step 5).

4. Entry component — owns its own state

In a sessioned integration this is the verification prompt (a banner or card) that offers "Verify to unlock". In a standalone integration it's whatever button or CTA triggers verification (an 18+ access gate, a download button, a social sign-up step). Either way:

  • Fetch the URL on-demand when the modal opens — never pre-fetch on parent mount. Pre-fetched URLs go stale across sessions and cause multiple simultaneous fetches.
  • verifyId is useState, NOT useRef. The polling useEffect depends on it, and useRef mutations do not trigger the effect to re-run.
  • Use a cancelled flag in the useEffect cleanup for in-flight request cancellation. Do not add a fetchingRef guard — it blocks re-fetch on modal close-then-reopen.
  • The create-verification call is server-side. The browser must never see the API key.

5. Detect completion — three strategies in parallel

Modal iframe completion can be detected via postMessage, but this is a hint, not authoritative. Always back it with a server poll.

When the experience is in a new tab (fallback when iframes are blocked, for example some enterprise setups), use all three:

  1. Server poll of GET /age-verification/get-status?id=<id> every 3–5 seconds.
  2. BroadcastChannel from the k-ID return page back to the app. The return page sends a message; the app listens at the top level (same place as the polling loop).
  3. Visibility wake — poll immediately on visibilitychange and focus when the user returns to the app tab.

get-status uses query param id, NOT verificationId. The wrong param is rejected with 400 and an empty body.

6. Post-verify — act on the result, race-free

On completion the handler is always idempotent and closes the modal before acting on the result:

const handleSuccess = useCallback(async () => {
  if (resolvedRef.current) return;
  resolvedRef.current = true;
  setModalOpen(false);
  await onVerified();
}, [onVerified]);

What onVerified does depends on the pattern.

  • Pattern 1 — Standalone AgeKit+. onVerified reads the result (PASS / FAIL) from /age-verification/get-status?id=<id> or from the webhook payload your backend received, and either grants access or shows the failure state. There is no session to refresh.
  • Patterns 2 and 3 — Sessioned. onVerified calls GET /session/get once and adopts the updated session in a single step. Do not split this into two calls (for example a local "clear unverified flag" followed by a network "refresh session") — they race and the stale state wins a meaningful fraction of the time.

Gotchas

  • sessionId is optional, not required. Include it in Patterns 2 and 3 where a k-ID session exists; omit it in Pattern 1 (standalone AgeKit+). Many integrators assume it's required because the sessioned examples always show it — forcing a session into a standalone flow adds complexity for nothing.
  • Body for perform-access-age-verification is nested. All age inputs go under criteria (criteria.ageCategory, criteria.minimumAge). Top-level fields are rejected with 400 in every pattern.
  • verifiedAgeThreshold permissions are never GUARDIAN-managed. Parental consent cannot unlock a Brazil loot-box or a UK OSA 18+ feature — only a successful age assurance can. A consent challenge against a threshold permission resolves PASS but the permission stays locked, so route threshold permissions through /age-verification/perform-access-age-verification.
  • Verification method selection is server-driven. Do not hardcode "show facial age estimation" or "show ConnectID". The url returned by /age-verification/perform-access-age-verification already reflects the correct method for the jurisdiction and Compliance Studio configuration. Client code just renders the iframe.
  • "Highly effective age assurance" (UK OSA) is a Compliance Studio setting, not a client flag. The skill never hardcodes the OSA standard. The verification URL is generated to meet it when the jurisdiction is GB and the permission requires it. Do not filter methods client-side.
  • /age-verification/get-status query param is id, not verificationId. Wrong param returns 400 with an empty body.
  • window.open / target="_blank" breaks iOS Safari. Always prefer an iframe modal. Treat new-tab as an enterprise fallback, not the default. If you must use a browser-based flow (no iframe available), set options.redirectUrl so k-ID redirects back with verificationId and result query params.
  • verifyId MUST be useState, not useRef. Polling useEffect lists it in its dependency array; refs do not cause effects to re-run.
  • Do not guard the URL fetch with fetchingRef. It blocks legitimate re-fetches when the modal closes and reopens. Use a cancelled flag in cleanup instead.
  • postMessage is a hint, not authoritative. Always back with a server poll against /age-verification/get-status or a webhook.
  • Six known new-tab detection failures, each seen in real integrations: popup blocked; tab backgrounded; BroadcastChannel unsupported; window.open returned null; the return page doesn't load before close; the user closes the tab before the webhook fires. The three-strategy approach in step 5 covers all six.
  • One post-verify path, not two. Whether you're refreshing a session or acting on a standalone result, run exactly one handler after the idempotency guard fires — never two that could race.
  • resolvedRef.current idempotency guard. The three detection strategies can all fire at once. Without the guard, you attempt the post-verify step twice and end up with an inconsistent UI.
  • Close the modal BEFORE acting on the result. Acting first can cause the UI to redraw with stale state momentarily.
  • Don't pre-generate verifications. Create the verification only when the user takes the action that starts the flow — never on page load or modal mount-and-hide. Pre-generation wastes verification counts and breaks k-ID's usage metrics. See best practices.
  • Verification URLs expire after 2 weeks. If a user abandons and returns later, check status first; if the URL is expired, create a new one rather than reusing the stale link.
  • Zero padding around the iframe. The k-ID widget owns the visual space. Padding breaks layout and looks off.
  • shortUrl is opaque. Display it as returned. Don't rebuild the URL, don't re-shorten, don't decode assumed query params.

Defaults

  • Modal: 480px wide, min(720px, calc(100vh - 48px)) tall, 16px radius.
  • Toolbar: thin (32–40px), title + literal x close button.
  • Iframe: flex: 1, border: none, no background.
  • Poll interval: 3–5 seconds while the modal is open.
  • Wake triggers: visibilitychange, focus.
  • Retry on transient /get-status failures: 2 retries, 2s apart.

Validation step

Confirm end-to-end on real devices.

Pattern 1 — Standalone AgeKit+

  1. Calling /age-verification/perform-access-age-verification with the target jurisdiction and criteria (no sessionId) returns id, url, shortUrl.
  2. Loading url in an iframe lets the user complete verification end-to-end.
  3. After completion, either the webhook your backend registered fires, or GET /age-verification/get-status?id=<id> returns the final PASS or FAIL, and the product acts on it correctly.
  4. The API key is never visible in browser network traffic.

Patterns 2 and 3 — Sessioned

  1. Complete verification via iframe modal — the gated feature unlocks within 10 seconds of the k-ID widget confirming success.
  2. Force the new-tab fallback (iframe blocked by CSP in a test environment) — feature still unlocks via poll or BroadcastChannel.
  3. Backgrounding the app tab during verification and returning — the feature unlocks immediately on focus (visibility wake fires).
  4. GET /session/get after verification returns the same sessionId with ageVerificationStatus: "verified" and the full permissions set.
  5. Hitting the verify banner twice rapidly — exactly one modal opens (idempotency guard works).

Next steps

Canonical references:

Keep looking

Skills are one crate of 608,475. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.