More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    sectionsight

    1

    Privacy-first website analytics your agent builds into your site, then proves in a real browser.

    Free

    1 useSecurity scanned
    sectionsight

    sectionsight

    Example session with this skill installed

    Add privacy-first analytics to this site. I want to know which sections people actually read, get a Slack message for interesting visits, and keep PostHog replay only for people who accept.

    • Read your context and instructions
    • Compiled the sectionsight
    • Generated the UI component

    ## Summary Analytics are in, with **consent model B**: totals for everyone with no cookies, per-visit rows and PostHog replay only after Accept. Everything ran through the browser check below. ## What I asked you, and your answers | Decision | Answer | |---|---| | Consent model | B (banner, ids only after Accept) | | Where the data lives | your MongoDB, `analytics` database | | PostHog replay | yes, after consent, EU host, inputs masked | | Notifications | Slack `#visits`, health alerts to `#ops` | | Retention | totals 13 months, visit rows 90 days, Slack lines 30 days | | Private paths | `/account`, `/reset`, `/invite` | ## What I added - `public/k.js` + `data-*` settings in `BaseLayout.astro` (sections are your `<section>`s under `<main>`) - `src/pages/api/k.ts`: the same-origin collector (`collector.mjs`) - `public/consent.js` + `consent.css`: banner, "Privacy settings" link in the footer - `posthog-guard.mjs`: starts PostHog only after Accept, stops it on withdrawal - `digest.mjs` (Mondays 08:00 Europe/Prague), `health.mjs`, `retention.mjs` - `ANALYTICS_DECISIONS.md` and a draft section for your privacy page ## The check (`node scripts/verify.mjs https://localhost:4321 --privacy-page /privacy`) ``` PASS before consent: no cookies PASS before consent: nothing stored in the browser PASS before consent: no PostHog request PASS before consent: reports carry no ids (mode consent) PASS reports and events never carry the probe e-mail or token PASS a refusal is remembered (no banner on the next page) PASS after consent: PostHog sends events PASS withdrawal: PostHog sends nothing more, in any tab PASS Global Privacy Control: nothing is sent PASS the privacy page names every service the browser sends data to ... PASSED (26 passed, 1 skipped) ``` ## Next steps 1. Read and adjust the privacy text draft (it is a draft, not legal advice). 2. Set `SLACK_WEBHOOK_URL` and `POSTHOG_KEY` in your host's env, then deploy. 3. Run the same check against production once it is live.

    consent.css

    TSX · React component

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Ask your coding agent to "add privacy-first analytics to this site". sectionsight asks you a few decisions, builds a cookieless setup into your own site, and then checks it with a headless browser. What you get: - Time actually spent on each page section and whether it was read, scroll depth, CTA and goal clicks, dead and rage clicks, 404s, JS errors and Core Web Vitals. No cookies, no ids. - A consent banner designed around the EDPB Cookie Banner Taskforce report and CNIL guidance. PostHog session replay is optional and only runs after consent. - A Slack or e-mail line per notable visit, a weekly digest, AI referral and crawler counts, health alerts, enforced retention, and self-service deletion of a visitor's data. - A draft of the privacy-page text. The check (scripts/verify.mjs, Playwright) proves the claims: nothing stored and no ids before consent, no e-mail or token in any request, withdrawal works across tabs, and, with --privacy-page, that the privacy page names every service the browser sends data to. Works on plain HTML, Astro, Next.js, React SPAs, Cloudflare and Node. Installs as a plugin in Claude Code, Codex and Gemini CLI. Not legal advice: the skill never claims compliance. Open source (MIT). Latest version, issues and changelog: https://github.com/Majny/sectionsight

    Needs a same-origin POST endpoint (a server, function or worker); a static host with nothing behind it is not enough. The browser check needs Node 20+, Playwright and Chromium. Self-service deletion is not available on Cloudflare Analytics Engine. The privacy-page text is a draft for the owner to review, not legal advice.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Slack
    Posthog
    esbuild
    Node.js
    playwright

    Permissions

    Terminal / Shell
    Read Files
    Browser
    Network Access
    Environment Variables
    Write Files

    Allowed Hosts

    eu.posthog.com
    slack.com

    Writes the analytics files into your project (only after you answer its questions). Shell and browser: runs the Playwright check against your site. Network: installs Playwright and Chromium for the check; your site talks to PostHog or Slack only if you choose them. Environment variables: the Slack webhook and PostHog key are read at runtime and never written into the repo.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    It runs on your own site and your own storage, with no cookies and no ids before consent. It measures what people read (seconds per page section), not just page views, and a headless-browser check shows what is and is not sent. It does not claim legal compliance; it gives you evidence to show your reviewer or lawyer.

    Plain HTML, Astro, Next.js, React SPAs, Cloudflare Workers and Node, anything that can take a same-origin POST. It installs as a plugin in Claude Code, Codex and Gemini CLI, and any agent that loads SKILL.md can use the folder.

    The agent reads your project without changing anything, asks you a few decisions (consent model, where data lives, replay, notifications, retention, private paths), builds the setup, and runs scripts/verify.mjs against your site.

    The tracker (k.js), the collector and its stores, the consent banner, the PostHog guard, Slack/e-mail notifications, the weekly digest, health alerts, retention and deletion scripts, reference docs for each part, and the Playwright check. MIT licensed, free.

    Yes. You pick the consent model (totals only, or per-visit rows after Accept), retention for totals, rows and Slack lines, and which paths are never measured. Your answers are written into ANALYTICS_DECISIONS.md.

    With the banner model, only totals with no ids: no cookies, nothing in browser storage, no PostHog request. The check fails if any of that changes.

    New in Testing & Debugging