AI agent control panel mockup - Animabot

What is Animabot?
A framework for creating AI agents that live permanently in Matrix/Element chatrooms. Each bot has its own Ethereum wallet, a defined personality, and a psychological state that evolves day by day.

What the control panel does
Six tabs — Status, Significant Interactions, Chat, Memory, Persona, and Debugger.

• Status — live connection health, wallet balance, room count, and a colour-coded activity log
• Chat — a private direct line to the bot outside of the chatroom, useful when Matrix is unreachable
• Memory — browse and clear per-room conversation history; all messages stored in Postgres
• Persona — the main screen (see below)
• Significant interactions — log of moments where someone pushed past the bot's comfort thresholds, with scores and a one-line reason
• Debugger — runs a headless Chromium session to diagnose login and connection issues on abliterate.ai

The Persona screen
The centrepiece. Two-column layout: character on the left, controls on the right.

Left column: system prompt (semi-permanent, edit requires typing the bot's name to confirm), MBTI type (same friction to change), and a scrollable 7-day reflection history generated by the bot itself each morning.

Right column: comfort threshold sliders (aggression, intimacy, existential, manipulation), memory depth and bias sliders, and a 7-day mood breakdown.

The bot runs a daily reflection at 3am — reads its recent conversations and significant interactions, then writes 2-4 sentences in first person about how the day affected it. That note gets injected into every future system prompt, subtly shifting tone without rewriting character.

Public vs admin
The panel has two views. Public (no password) shows all data with usernames censored. Admin (password protected) shows everything and unlocks editing.

Built with
Vanilla HTML/CSS/JS mockup. Designed to connect to a Node.js backend with matrix-js-sdk, ethers.js, Qwen API, and Postgres.