Hermes Agent 0.6.0 Multi-Agent Profiles Update!
Hermes Agent 0.6.0 launches with multi-agent profiles, allowing users to run multiple isolated AI agents simultaneously from a single installation. This open-source update addresses a major weakness by enabling parallel agent workflows with separate configurations, memory, and API keys.
Summary
Hermes Agent 0.6.0 represents a significant milestone for the open-source AI agent platform developed by Nose Research. The update's flagship feature is multi-agent profiles, which solves the platform's biggest limitation - previously only allowing one agent to run at a time. Users can now create multiple fully isolated Hermes instances from a single installation, each with separate configurations, API keys, memory pools, sessions, and skills. The system includes safety measures like token locks to prevent credential conflicts between profiles, and supports profile export/import for team collaboration.
Beyond profiles, version 0.6 introduces MCP (Model Context Protocol) server mode, enabling Hermes to integrate with tools like Claude Desktop, Cursor, and VS Code. The update adds fallback provider chains for improved reliability - when a primary AI provider fails, the system automatically switches to backup providers without interrupting workflows. New platform integrations include Fau/Lark and WeChat for Chinese enterprise users, Slack multi-workspace support, and Telegram webhook mode for better performance. The release also includes an expanded migration guide for OpenClaw users, making it easier to switch platforms. With 18,200 GitHub stars and 95 pull requests in just 2 days, the community momentum suggests Hermes is becoming a serious production-ready tool for automated workflows.
Key Insights
- Hermes Agent's single biggest weakness before version 0.6 was only being able to run one agent at a time, forcing users to choose between different use cases like customer support, content drafting, or research
- The new profiles feature creates completely isolated Hermes instances where each profile gets its own command structure, memory, and gateway service with zero crossover between different agents
- Token locks provide production-ready safety by blocking any attempts for multiple profiles to use the same bot credentials and providing clear error messages about which profile owns each credential
- MCP server mode allows Hermes conversations and history to become accessible from other AI tools like Claude Desktop, Cursor, and VS Code through a standardized protocol
- Fallback provider chains enable automatic switching between AI providers when the primary service fails, transforming Hermes from a toy into a production-ready tool that maintains workflow continuity
Topics
Full transcript available for MurmurCast members
Sign Up to Access