Owner profile
Store stable preferences: name, timezone, working style, language preference, and recurring constraints that affect most answers.
A useful Telegram assistant needs stable context, not unlimited memory.
Use workspace files and project notes to keep durable preferences available while avoiding noisy, stale, or sensitive context in every request.
The best memory setup is small enough to review and specific enough to improve everyday answers.
Store stable preferences: name, timezone, working style, language preference, and recurring constraints that affect most answers.
Define tone, boundaries, output style, and when the assistant should act versus ask for confirmation.
Keep current goals, active repos, launch status, deployment notes, and marketing decisions in project-specific notes.
Record local setup details that are safe to store: preferred channels, repo paths, camera names, or non-secret service names.
Do not put tokens, passwords, private keys, or sensitive customer data in plain workspace notes.
Prune stale assumptions after launches, migrations, pricing changes, or tool upgrades so the assistant does not repeat old advice.
Do not treat memory as a dumping ground. Long logs, old drafts, secrets, temporary debugging output, and vague motivational text make the assistant less reliable. Store the decision, not every path that led to it.
When the assistant knows the notification rules, project state, and safe next actions, daily checks can verify evidence, make low-risk improvements, and stay quiet unless the owner actually needs to decide something.
The OpenClaw Telegram Assistant Launch Kit includes persona and workspace templates so a private assistant starts with useful behavior, privacy boundaries, and practical operating rules instead of a blank prompt.