One human, one agent, one set of keys — yours. No SaaS account, no shared infrastructure, no strangers in your data.
You bring your own keys and your own agent instance. Each user's costs, rate limits, and conversation history live with that user — never pooled.
Your agent ignores everyone except you. There's no public address, no admin backdoor, no shared instance that could leak across accounts.
Identity, voice, memory, and rules are per-user. The agent learns who you are over time and adapts — without exposing any of it to anyone else.
Create an account and give your agent a name. If you already have a Celindor user, link it with a claim code instead.
Paste an API key or Claude Code OAuth token. We encrypt it before it touches the database — you pay Anthropic directly, nobody else can spend on your account.
Your agent comes online within seconds. It remembers facts about you, runs tools on your behalf, and asks before doing anything risky.
Whatever Anthropic charges you for tokens, plus optional OpenAI for embeddings. Celindor itself is free and open source — there's no Celindor subscription.
Even if someone finds a way to reach your agent, the gateway silently drops every message that doesn't come from your linked account.
In a Postgres database shared across users only at the row level — every table is keyed by your user id and isolated by row-level security. The agent runs on a single Linux box. Nothing leaves your record.
