Frona

Frona is a self-hosted AI assistant for people who want agents to carry out tasks through chat with controlled access to files, networks and accounts. Agents can work independently or delegate parts of a task to other agents.

• Task execution: Agents browse websites, search the web, run shell commands, Python or Node.js, and build applications, with your approval required before deployment.
• Access controls: Separate Linux sandboxes restrict each agent's files, network destinations and resource use, while isolated browser profiles keep users' sessions separate.
• Credential approval: Agents request access through 1Password, Bitwarden, HashiCorp Vault or KeePass, and you approve or decline; secrets stay out of agent memory and LLM requests.
• Persistent memory: Choose compact facts and notes or an ontology-backed knowledge graph with evidence, relationships and reusable playbooks, and inspect memories and consolidation history in the interface.
• Ongoing work: Schedule recurring tasks and monitors, or let agents pause until a reply or verification code arrives through Telegram, Slack, Discord, WhatsApp or Signal.

Frona runs on your infrastructure in an OCI container using Docker, Podman or another compatible runtime. Its application data lives on your servers, while model requests go directly to the LLM provider you choose.

MCP servers and skills extend agent capabilities. Browser automation uses Browserless, and web search supports SearXNG, Tavily or Brave Search. Optional Twilio integration provides outbound phone calls and SMS.