How is this different from Playwright MCP?
Playwright MCP wraps Playwright's CDP-based architecture with an MCP layer — it's still async-by-construction and bottlenecked by the marshalling tax between processes. Alethia is a different architecture entirely: the executor and the DOM live in the same V8 isolate, no CDP, no async race window. That's where the 2–5× per-flow speedup vs Playwright MCP comes from (and up to 50× vs the Playwright CLI path that most CI suites run today). It also ships with a fail-closed safety policy gate (EA1) and a cryptographically chained audit trail. Features Playwright doesn't provide. Both tools are useful; they serve different jobs.
What does this cost? What about after the eval?
The MCP bridge (@vitronai/alethia) is MIT-licensed and free,
permanently. The runtime is free to evaluate. Install it and use it
today, no approval required. Production and commercial use requires a
separate license; pricing will publish publicly when the patent grants.
Email gatekeeper@vitron.ai to discuss licensing.
Can I self-host this in my enterprise environment?
Yes. That's the entire architecture. Alethia runs entirely on the customer's
machine, with no cloud dependency, no telemetry by default, and no data
leaving the local network. The runtime listens on a loopback port
(127.0.0.1:47432) that's never exposed externally. For regulated
environments (financial services, healthcare, federal civilian agencies,
defense contractors), this is the deployment story we're built around.
Does this work in air-gapped or classified environments?
Yes. After the one-time runtime download (which can be done on a connected machine and transferred via approved media), Alethia runs fully offline with zero network dependencies. No telemetry, no license-phone-home, no cloud calls on any code path. The runtime is a single binary with no package manager, no dependency resolution, no external fetch at runtime. The audit trail is generated locally, signed locally, and never transmitted. U.S. patent, U.S. company, no foreign filings or operational dependencies.
What data leaves my machine?
None, by default. Alethia is local-first by architecture: telemetry is off by default, the runtime listens on loopback only, and there are no cloud calls on the execution hot path. The auto-installer fetches an Ed25519-signed runtime binary from GitHub Releases exactly once, on first install — that's the only network call the system makes. After install, nothing reaches out to anything.
Does it work with my framework? React / Vue / Next.js / Svelte?
Yes. Alethia drives any web application that runs in a browser, regardless of framework. It interacts with the rendered DOM, not with framework internals, so React, Vue, Svelte, Solid, Next.js, Remix, Nuxt, SvelteKit, Astro, vanilla HTML: they all work the same way. The selector engine uses a multi-strategy fusion (test-id, aria, role+name, text, structural fallback) so it's resilient to refactors that would break a CSS-selector-only tool.
Can I point Alethia at a production URL or staging environment?
No. Alethia only navigates to local origins — localhost,
127.0.0.1, file://, and private network
ranges. This is a compile-time constant, not a setting you can change.
Alethia is built for the local test loop: the app you're developing,
running on your machine. If you need to verify against your own
staging or production domain, contact
gatekeeper@vitron.ai — we can discuss custom builds
for named organizations.
Full posture: vitron.ai/safety.
What about CI? Can I run this headless?
Headless mode runs in any Node-capable environment, including cloud CI services and self-hosted runners. The bridge auto-installs the headless runtime on first call, so CI integration is the same as installing any other npm package. The audit pack each run produces is designed for CI consumption: JSON-formatted, exit-code-aware, and replayable from artifacts.