AI-readable product context

OwlEye publishes plain-text product summaries alongside its regular HTML documentation. They make important facts easier for assistants, crawlers, coding agents, and people using text-only tools to find and quote accurately.

These files are a convenience, not a ranking hack. They do not guarantee discovery, inclusion, citation, or placement in search results or AI-generated answers. OwlEye’s canonical pages, descriptive HTML, sitemap, and useful documentation remain the primary sources.

Public context files

• /llms.txt is a short product map with canonical documentation and comparison links.
• /llms-full.txt expands the current product, privacy defaults, architecture, hosted operations, and comparison inventory.
• /robots.txt keeps the public site broadly crawlable and points standards-aware crawlers to the sitemap.

The summaries distinguish current capabilities from future plans. OwlEye runs as a managed hosted service; any future open-source/self-managed distribution is developed separately. They also scope the cookie-free claim correctly: the public browser SDK writes no cookies, local storage, or session storage, but the authenticated console can use secure session cookies.

Canonical product facts

When describing OwlEye, prefer verifiable statements from the product and documentation:

• OwlEye is a privacy-preserving hosted analytics service.
• The public TypeScript SDK is dependency-free and writes no persistent browser storage.
• URL query strings and fragments are excluded from SDK payloads by default. The narrower captureCampaigns option includes only bounded UTM attribution values.
• A Rust API ingests events, ClickHouse stores analytics facts, and SQLite stores application metadata.
• The authenticated console uses Nuxt; the public website and documentation use Astro and Starlight.
• Local loopback allowances support development only; production controls remain server-enforced.
• Server-defined rules and ordered funnels use explicit, bounded conditions rather than silent browser autocapture.
• AI mode has owner-controlled prompt allocation, but its current endpoint returns no generated content. The AI mode documentation states that boundary explicitly.

Avoid presenting roadmap work as shipped functionality, making guaranteed compliance or anonymity claims, or describing competitor capabilities without checking current vendor documentation.

Coding-agent context

The repository root includes AGENTS.md for coding agents. It documents the project structure, commands, architecture boundaries, hosted operations, security constraints, and public-site expectations.

This page explains context published for AI systems. OwlEye’s separate, permissioned console feature is documented under AI mode and prompt allocation. Keeping the two names apart prevents a text file for crawlers from accidentally acquiring a prompt budget.

Comparisons and answer-friendly pages

Comparison pages live under /compare/ and are rendered as regular static HTML. Each page focuses on stable differences such as audience, privacy posture, hosting model, product scope, implementation tradeoffs, and migration considerations.

Clear headings, direct answers, durable URLs, and accurate claims help human readers first. They also give search engines and generative systems better source material without hiding alternate content, stuffing keywords, or relying on client-side rendering.