CLItrail Docs

Add CLItrail to Sphinx

Where the CLItrail tag goes in Sphinx, in its own idiom, with the consent, CSP and client-side routing details and how to check it works.

Written for Sphinx 9 (script attributes since 1.8)

#Where the tag goes

Edit conf.py. Put the tag on every page that shows your install command; site-wide is best, after your existing analytics tags.

html_js_files = [
    ('https://YOUR_SERVICE/browser.js', {'defer': 'defer', 'data-website': 'YOUR_WEBSITE_ID'}),
]

Use the loader in place of the tag: nothing from CLItrail loads until startClitrail() runs. Load the loader before your consent manager's script and call startClitrail() from the manager's accept callback (most managers also run it on later visits once a choice is stored; check yours). Use the marketing category when installs go to Google Ads, Meta, TikTok or X Ads, and the analytics category when they only reach GA4 or your webhooks.

Consent loader

// Loads CLItrail only after the visitor agrees. Call startClitrail() from your
// consent manager's "accepted" callback; nothing is fetched or stored before that.
window.startClitrail = () => {
  if (document.getElementById('clitrail')) return;
  const tag = document.createElement('script');
  tag.id = 'clitrail';
  tag.src = 'https://YOUR_SERVICE/browser.js';
  tag.dataset.website = 'YOUR_WEBSITE_ID';
  document.head.append(tag);
};
html_static_path = ['_static']
html_js_files = [
    ('clitrail-consent.js', {'defer': 'defer'}),
]

#Client-side routing

Every page is a full page load, so the tag records each page by itself. Nothing else to add.

#Content Security Policy

The build is static files, so the policy lives in your host's header settings (for example _headers on Cloudflare Pages or Netlify, or vercel.json): add https://YOUR_SERVICE to script-src and connect-src, and blob: to worker-src (only Safari before 26 needs that one).

script-src 'self' https://YOUR_SERVICE;
connect-src 'self' https://YOUR_SERVICE;
worker-src 'self' blob:;

#Verify it works

  1. Build with make html (or sphinx-build) and serve the output folder, or publish.
  2. Open a page with the tag in Chrome or Firefox (accept analytics first if you gate CLItrail on consent), open the developer console and run await InstallAttribution.ready. It resolves to { ok: true, urlId, hasAnalyticsContext, expiresAt }.
  3. In the CLItrail dashboard, open Visits & identities: the visit is listed.
  4. Anything else names the cause: disabled (no consent yet, or data-enabled="false"), opfs_unavailable (the page is not served over https or from localhost), Error (the service refused the visit: add the page's origin, including a development origin such as http://localhost:3000, under Settings → Additional domains). If InstallAttribution is undefined, the tag did not load: check the Network tab and your Content-Security-Policy.

Then run your installer on the same computer and check Install events, or run the hook with --doctor.

#Notes

Sphinx accepts a full URL in html_js_files and writes the dictionary as the tag's attributes.

Sphinx documentation: www.sphinx-doc.org/en/master/usage/configuration.html

Using other install paths or destinations? The setup generator puts this snippet together with your hooks and destination checklist.