We don't profile you.
Vibestrate is open-source software maintained by its author. No ad networks, no profiles of individuals, no selling or sharing of personal data. This site collects only a little anonymous analytics, and the software runs on your own machine. Here's everything we touch.
This site's analytics
We use two cookieless tools, both run at the edge - nothing is stored on your device:
- Cloudflare Web Analytics - aggregate page views, country, referrer, and browser family. No cookies, no stored IP, no fingerprinting.
- A first-party referral beacon to stats.vibestrate.com that, once per session, records the page you landed on, the channel you came from (search, social, an AI tool like ChatGPT, or direct), any utm_* tags, and a coarse country and device class - no IP, no identifier, no account.
Beacon data lives in a Cloudflare D1 database we run and is deleted after 180 days. A tracker-blocker (or blocking stats.vibestrate.com) stops both - the site works fine without them. No Google/Facebook/LinkedIn/Hotjar scripts, no ads, no forms, no signup; fonts are self-hosted.
The flow hub
Vibestrate has a community flow hub - a place to find and install shared "flows" (run recipes). It's the one part of the software that talks to a server we run, and it only does so when you ask it to. When you search the hub or install a flow, the software calls our hub at vibestrate.com/api/hub. That request:
- Exposes your IP and User-Agent to the hub and to Cloudflare, like any web request. We use it only to return the flow you asked for.
- Carries what you searched for or the flow you chose to install - never your code, prompts, or other projects.
- Needs no cookie and no account to browse or install.
This never happens in the background. Publishing a flow (where enabled) is a separate action: it needs you to sign in, and your publisher handle and the flow you publish become public. Flows you install are third-party content - inspect them before running (see our Terms).
The software stays on your machine
In the current open-source release, which you can read in full yourself:
- Your API keys never reach us. They live in the vendor CLIs (claude, codex, gemini, ollama), the same place they already were. Vibestrate just spawns those CLIs and reads their output.
- No model proxy. Your prompts and the model's responses go straight between your CLI and the provider. We never see, log, or forward your code, prompts, or output.
- The cost ledger is local. Token and cost figures are estimated on your machine and saved to a local SQLite file - never uploaded or aggregated. We don't know who runs Vibestrate, how often, or at what cost.
- No background calls. Aside from the flow hub above (search, install, and publish - only when you ask) and your package manager's version check on install, the software makes no outbound calls.
Your data and your rights
Everything we keep is aggregate and not tied to your identity, so - aside from anything you choose to publish to the hub under a handle - there's no profile of you for us to access, export, or delete. We keep this little on purpose. Depending on where you live, your local laws may give you rights over personal data; ask and we'll do what applies.
To raise anything privacy-related, open a GitHub Discussion, or email [email protected] where a direct contact is legally required.
Hosting and changes
Cloudflare hosts this site and the services we run, and may process the little data above (and the connection metadata every website sees) on infrastructure outside your country, including the United States. We may update this page; with no accounts, the only notice is here (with a new date) and the project's GitHub. This site and the software are not directed to children under 13.
