HQ is a self-hosted, Discord-styled project hub with LLM-powered chat, parallel agent tasks, per-channel memory, and full multi-provider AI routing — no SaaS lock-in.
Built for power users who want full control over their AI stack.
Discord-style channels per topic — #research, #backend, #agents. Each channel has its own AI context and system prompt.
Spawn up to 10 background AI agents simultaneously. Each runs an autonomous tool-use loop with web search, step tracking, and cancellation.
Groq, Gemini, Together AI, Claude OAuth, OpenRouter, Cerebras, SambaNova, Mistral, Ollama. Account Stacking: pool multiple free-tier keys per provider — HQ treats them as a single high-bandwidth stream and jumps to the next healthy key the moment one is throttled.
Send one message to two models simultaneously and compare responses side-by-side. Great for evaluating outputs or getting a second opinion.
Never repeat yourself. HQ's Handoff Protocol acts as a "Project Brain" that stays updated in real-time — distilling facts and decisions from your chat into a persistent state so the AI (and your team) always knows exactly where the project stands.
Granular permissions system. Admins manage rate limits, restrict model access, view audit logs, and manage user roles from a 4-tab admin panel.
Every AI response has access to web tools: fetch any URL via Jina Reader, search via Tavily or DuckDuckGo. Autonomous 5-step tool loop.
Real-time provider health indicator in the composer. Green = healthy, amber = rotating to fallback key, red = all keys rate-limited.
One slash command generates a complete handoff document summarizing the channel state, decisions made, and next steps — ready for a new model or collaborator.
Escape context lock-in. Import your entire Claude or ChatGPT history — conversations, projects, and memories — into HQ in one click. Your past context, finally portable.
OSS Core is free forever. Pro Command License unlocks Account Stacking, unlimited agents, and history import.
You always pay AI providers directly. Typical Groq costs: ~$0.06/M tokens — a team of 5 spends ≈ $2–5/mo total.
Try the live demo instantly — no account setup required.