Show HN: Dexto – Connect your AI Agents with real-world tools and data

6 months ago

Dexto is an all-in-one toolkit for building agentic applications that convert natural language into real-world actions.
It serves as a universal intelligence layer for creating collaborative, context-aware AI agents and applications.
Dexto combines a configuration-driven framework, robust runtime, and seamless developer experience for easy agent development.
Key features include a YAML-based framework for defining agent behavior, runtime execution with orchestration, and native support for CLI, Web, APIs, and a TypeScript SDK.
Dexto supports autonomous agents, digital companions, multi-agent systems, and agent-as-a-service applications.
It includes over 50 LLMs, multimodal support, persistent sessions, and pluggable storage options.
Dexto can be installed via npm or built from source, and it offers various modes of interaction including Web UI, CLI, and server modes.
Pre-built agent recipes are available for common use cases such as coding, image generation, podcast creation, and more.
Dexto agents are modular, composable, and portable, allowing for easy integration and extension via Model Context Protocol (MCP) servers.
The toolkit includes advanced features like human-in-the-loop approval policies, observability, and flexible deployment options (local, cloud, or hybrid).

Hasty Briefsbeta