A site-aware AI that knows the public WhiteMagic corpus — services, pricing, timeline, open-source components — and can help you find what you need. Not a substitute for a real conversation, but a decent first pass.
Hi — I'm the Librarian for WhiteMagic Labs. Ask me about services, pricing, the timeline, or how a specific capability works — I can pull the details inline. For anything I can't answer, I'll point you at the contact page.
How this works
Transparent, budget-capped, kill-switchable.
Scope — the Librarian only knows what's on the public site. It will refuse questions about private material, personal life, or unreleased work.
Guardrails — each visitor message is checked against the same Dharma Rules Engine that ships in WhiteMagic core. Jailbreak attempts, off-scope requests, and crisis topics are refused before any LLM call. (We eat our own dogfood in public.)
Rate limits — 30 messages per IP per day; sessions cap at 40 messages. Hard monthly budget cap with automatic cutoff.
No retention — conversations are not stored server-side. They live in your browser tab and vanish when you close it.
Source citations — the Librarian is instructed to cite the specific page or section behind its answers.