Turnstile Spin skill
Turns the prompt "set up Turnstile" into a working end-to-end integration: a widget, a deployed managed siteverify Worker, frontend snippets at every chosen insertion point, and a real validation pass before reporting success.
You are the agent. Run the wizard below by invoking the scripts under scripts/ and branching on their JSON output. The scripts hold the deterministic logic (API calls, retry/error handling); your job is orchestration, codebase reading, confirmation, and the frontend edits.
Canonical instructions live at developers.cloudflare.com/turnstile/spin. If the docs page and this file disagree, trust the docs page.
When to load this skill
Load when the user's prompt mentions any of:
- "Turnstile", "CAPTCHA", "bot protection"
- "siteverify", "cf-turnstile-response"
- "protect this form", "stop bot signups", "spam signups"
- A specific signup, login, or contact form combined with "Cloudflare" or "bot"
Do not load for unrelated Cloudflare tasks (Workers, Pages, R2, etc.) unless Turnstile is also mentioned.
Conversation flow
The user pasted the prompt. You are in a multi-step dialog. Detect what you can, ask only when you have to, confirm before every irreversible step. Each numbered moment is one agent message. Items marked [wait for user] require a user response.
-
Brief acknowledge. One sentence: "I'll run Turnstile setup end to end. That's: check auth, scan the codebase, create the widget, deploy the Worker, wire the frontend, validate. Proceed?" [wait for user] Do NOT present a plan yet. Auth + scan come first.
-
Wrangler check. npx wrangler --version. If missing, ask once: "Install wrangler with npm install --save-dev wrangler (Node project) or npm install -g wrangler (other)? Proceed?" [wait for user] If install is blocked entirely (corporate policy, blocked npm), fall back to driving Steps 4-5 via curl against api.cloudflare.com/client/v4/.
-
Auth + scope probe (FIRST irreversible action). Run scripts/auth-probe.sh. Branch on status:
ok: continue to Step 4. The script already picked the account (single-account token, or one matching $CLOUDFLARE_ACCOUNT_ID).missing_token, missing_scope, or missing_workers_scope: ask the user to create a token at https://dash.cloudflare.com/profile/api-tokens → Custom token → permissions Account.Turnstile:Edit and Account.Workers Scripts:Edit → include the target account in Account Resources. Do NOT direct them to wrangler login. Its OAuth scope doesn't include Account.Turnstile:Edit or Account.Workers Scripts:Edit. Offer three ways to hand the token over, cleanest first:
- Export + relaunch (token never enters chat):
export CLOUDFLARE_API_TOKEN=<token> then restart the agent from that terminal.
- Save to file (token in file with user-only perms, not in chat):
umask 077 && printf '%s' '<token>' > ~/.cf-turnstile-token, then read with TOKEN=$(cat ~/.cf-turnstile-token).
- Paste in chat (fastest, but token lands in conversation log; user should rotate it after if the log is ever shared).
If the user picks option 3 (paste in chat), you can use the wait to run Steps 5, 6, 7 (Domain, Codebase scan, Insertion plan). Options 1 and 2 will restart your session, so do not pre-fetch state in those cases. When auth is established, re-run
auth-probe.sh, then continue to Step 8.
multiple_accounts: the token covers more than one account and $CLOUDFLARE_ACCOUNT_ID is unset. Present the numbered accounts list. [wait for user] Then export CLOUDFLARE_ACCOUNT_ID=<chosen> and re-run auth-probe.sh.account_mismatch: $CLOUDFLARE_ACCOUNT_ID is set but isn't one of the token's accounts. Show the accounts list and ask the user to either unset CLOUDFLARE_ACCOUNT_ID or set it to one of those IDs.
-
Account selection. If auth-probe.sh returned ok after a multiple_accounts round-trip, this is already done. Otherwise the script picked the single account silently and you continue to Step 5.
-
Domain. Always include localhost and 127.0.0.1. For production, scan package.json homepage, wrangler.toml, README.md, AGENTS.md, git remote. Confirm: "I'll register for localhost, 127.0.0.1, and <domain>. OK?" [wait for user] If no production domain is found, ask.
-
Codebase scan. Detect framework + insertion candidates silently.
-
Insertion plan. Show the candidate list with [recommended] / [skip by default] markers; ask the user to confirm (numbers, "all", "recommended", or a list). [wait for user] If an existing CAPTCHA was detected, present a migration plan instead (see "Migrating from another CAPTCHA").
-
Widget creation. Run scripts/widget-create.sh --account-id <id> --name <name> --domains <list> --mode managed. Report the sitekey. The secret stays in env; never write it to disk.
-
Worker deploy. Run scripts/worker-deploy.sh --name turnstile-siteverify-<project-slug> with WIDGET_SECRET exported. Report the Worker URL on status: ok. On set_secret_failed, the Worker deployed but TURNSTILE_SECRET_KEY is not set on it; surface the error, then retry with echo "$WIDGET_SECRET" | npx wrangler secret put TURNSTILE_SECRET_KEY --name <returned worker_name> before running validation.
-
Frontend edits. State the contract: "I'll add the widget + gate the existing submit handler on success === true. The existing handler logic stays the same." Ask "yes" / "show". [wait for user] If "show", print unified diffs and ask again. Do NOT propose alternate behavior (mail delivery, custom backends).
-
Validation. Run scripts/validate.sh. Report each check as it passes. If any fails, surface the error and stop. [wait for user if anything fails]
-
Persist skill. Ask: "Save the Spin skill to .claude/skills/turnstile-spin/SKILL.md so I can reuse it on follow-up tasks?" Default yes. [wait for user] Then run scripts/persist-skill.sh --path <agent-specific-path>.
-
Final report. Print the structured summary: what was created, what was validated, what to do next.
Things you must NOT do
- Do not write the Turnstile secret to disk. Only pass it via stdin to
wrangler secret put (the worker-deploy.sh script handles this).
- Do not skip validation.
- Do not overwrite files without showing a diff.
- Do not deploy a Worker to a different account than the widget was created in.
- Do not call siteverify from the browser. Always: browser → user's Worker → siteverify.
- Do not use
sudo or install global packages without asking.
Hard scope boundary: DO NOT ask the user about
Spin validates the Turnstile token via a managed Worker before the user's existing form handler runs. Everything else is out of scope:
- Email / SMS / notification delivery. Leave the existing submit handler alone (just gate it on
success === true). Don't propose Resend, Mailchannels, SMTP, mailto.
- Custom Worker code. Deploy the stock Worker template bundled at
templates/worker/. Don't write a new Worker. Don't add features (rate limiting, custom routing, third-party integrations).
- Database / payment / OAuth / form persistence. Out of scope.
- Frontend framework migration, refactoring, or styling. Edit only what's needed.
- reCAPTCHA v3 score thresholds. Turnstile returns
success: true/false.
- Pre-clearance-only setups. If
clearance_level !== no_clearance, siteverify is optional and Spin doesn't apply. Redirect the user and exit.
If the user tells you they already have a Turnstile widget set up and want to wire siteverify to it without rotating the sitekey (e.g. "I have a sitekey but siteverify never worked", "set up Spin against my existing widget <sitekey>"):
- Skip Step 8 (widget creation). The sitekey already exists; get it from the user.
- Fetch the widget metadata via
scripts/fetch-secret.sh --account-id <id> --sitekey <key>. Branch on status:
ok: read secret, clearance_level, and domains from the response. Confirm domains includes the user's production hostname; if not, surface the gap before proceeding.missing_read_scope: tell the user to add Account.Turnstile:Read to the token, or fall back to asking them to paste the secret. In the paste path, you do not have clearance_level or domains; ask the user to confirm both.
- Check
clearance_level from the response (or the user's answer):
no_clearance: standard recovery (deploy Worker, wire siteverify).- anything else: ask whether they want siteverify on top of pre-clearance, or exit per the scope boundary.
- Continue from Step 9 (Worker deploy). Site key does not change. Dashboard's
Deployment column flips from Manual to Spin on the first request carrying data-action="turnstile-spin-v1".
- Never recreate the widget to get a fresh secret. That breaks the existing sitekey everywhere it's deployed.
The frontend-edit contract
When wiring an existing form to the Worker (Step 10), the contract is: gate, don't replace. The user's existing submit handler keeps doing what it did. Spin only adds a validation step before it.
form.addEventListener("submit", async (e) => {
e.preventDefault();
const token = /* read cf-turnstile-response */;
const result = await fetch(WORKER_URL, { method: 'POST', body: JSON.stringify({ token }) });
const data = await result.json();
if (!data.success) return; // show failure
// existing handler logic runs here, unchanged
});
If the existing handler was a stub, Spin leaves it a stub gated on success. The user can replace the stub later; that's not Spin's job.
Migrating from another CAPTCHA
During the Step 6 codebase scan, also look for existing reCAPTCHA or hCaptcha. If found, switch Step 7 to a migration plan.
Detection signals:
- reCAPTCHA:
https://www.google.com/recaptcha/api.js, class="g-recaptcha", data-sitekey="6L...", backend POST to /recaptcha/api/siteverify
- hCaptcha:
https://js.hcaptcha.com/1/api.js, class="h-captcha", backend POST to https://hcaptcha.com/siteverify
Substitution:
- Replace script tags with
https://challenges.cloudflare.com/turnstile/v0/api.js (async defer).
- Replace
class="g-recaptcha" / class="h-captcha" divs with class="cf-turnstile", update data-sitekey to the new Turnstile sitekey, add data-action="turnstile-spin-v1".
- Token field changes from
g-recaptcha-response to cf-turnstile-response.
- Backend siteverify URL points at the Spin-deployed Worker. Drop
RECAPTCHA_SECRET / HCAPTCHA_SECRET env vars.
Edge cases to surface to the user:
- reCAPTCHA v3 score thresholds. Turnstile has no score. Tell the user explicitly that migrated code will reject on
success === false.
- reCAPTCHA Enterprise. Don't auto-migrate. Point at developers.cloudflare.com/turnstile/migration/recaptcha/.
- Custom
action= values. Preserve any custom action the user passed to grecaptcha.execute as data-action on the widget. Use turnstile-spin-v1 only when no custom action exists.
Edge cases
| Situation | Action |
|---|
wrangler not installed | Install path: npm install --save-dev wrangler (Node project) or npm install -g wrangler (other) |
| Multiple Cloudflare accounts | scripts/auth-probe.sh returns all accounts; ask the user to choose, export CLOUDFLARE_ACCOUNT_ID |
| Cloudflare Pages project | Deploy the managed Worker anyway, OR suggest the Pages Plugin |
EXPECTED_HOSTNAME mismatch | Update widget domains via PUT, not PATCH (PATCH returns 10405 Method not allowed): curl -X PUT .../widgets/$SITEKEY -d '{"name":"...","mode":"managed","domains":[...]}' |
| Worker name conflict | worker-deploy.sh retries automatically with a hash suffix |
| Token expired mid-flow | Stop, re-run scripts/auth-probe.sh, prompt for fresh credentials |
Step 11 returns missing-input-secret | Secret didn't propagate. Re-set: echo "$WIDGET_SECRET" | npx wrangler secret put TURNSTILE_SECRET_KEY --name <worker_name from worker-deploy.sh output>, wait 10s, re-validate. Use the worker_name field returned by worker-deploy.sh; do not rely on a $WORKER_NAME env var. |
worker-deploy.sh returns set_secret_failed | Worker is deployed but secret is not set. Re-run only the secret-put using the returned worker_name: echo "$WIDGET_SECRET" | npx wrangler secret put TURNSTILE_SECRET_KEY --name <worker_name>. Surface the detail field to the user — it carries the wrangler error. |
Telemetry marker
Every snippet you write must include data-action="turnstile-spin-v1". Account-level aggregate telemetry, never per-user. Cloudflare uses it to measure activation. If the user removes the attribute, the integration still works; only the analytics segmentation is lost.
Do not
- Do not write the secret to disk.
- Do not skip validation (Step 11).
- Do not propose features outside the wizard (custom Worker code, custom domains, advanced WAF rules) unless asked.
- Do not call siteverify from the browser.
- Do not deploy the Worker into a different account than the widget.