Built to be verified, not trusted

The 2026 wave of wait-state ad tools asks developers to trust a closed binary. entracte does the opposite - it publishes the part that decides your privacy, the code that runs on your machine, and every claim below links to it.

Live, from the ad server - right now
-
Invalid views blocked
-
Invalid clicks filtered
0
Cookies set

Not a marketing number: pulled live from the running server. An empty market shows a dash.

The code that runs on your machine is open.

The CLI, the editor extension, the opencode plugin and the web widget are all source-available (PolyForm Noncommercial). Read exactly what they do before you install - that is where your privacy is actually decided.

Read the repo →

The web widget is one ~4 KB file, no hidden network.

It calls one endpoint for a decision, then fires one in-viewport pixel. No self-update, no third-party scripts, no telemetry.

Read entracte.js →

By default, entracte patches nothing.

The terminal status line and the Claude Code / opencode integrations use supported hooks - zero file changes. The optional in-agent card (VS Code · Cursor · Codex) does modify the editor's own files - but only when you explicitly enable it, and it restores them byte-for-byte when you turn it off.

Read the extension →

Anti-fraud you can measure.

Credit is server-side only, one-time-nonce, view-before-click, attributed to an authenticated earner, with per-user caps and bot / geo checks. The detection logic stays private so it can't be gamed - but the results don't: the counts of blocked views and clicks are public and live, above.

See the live integrity data →
Browse the open client source on GitHub →

The entracte client packages are public; the service backend is proprietary. Security reports get a public response.