October 28, 2025
Agents, outrage, and a license twist
Show HN: Dexto – Connect your AI Agents with real-world tools and data
Dexto promises push-button AI agents; HN erupts over 'not open source' claim
TLDR: Dexto launched a toolkit to build AI agents that can use files, browse, and remember, promising easy setup and local/cloud options. Hackers loved the demo but slammed the “open source” label (it’s ELv2) and demanded pricing, while memes mocked yet another “orchestrator for orchestrators.”
Dexto just landed with big talk: build AI helpers that can remember, browse, code, and click around the web from one prompt. The demo flex? “Make a snake game and open it in the browser.” Devs love the idea of easy, one-stop agents and the option to run them locally or in the cloud. The project boasts 50+ AI models, a slick web UI, and configuration in plain text. You can peek here: Dexto on GitHub.
But the celebration screeched when a top comment yelled: it says “open source,” yet the license is Elastic License 2.0 (ELv2) — not approved by the Open Source Initiative. Cue the courtroom drama: “Is this open source or just source-available?” The thread lit up with license-police energy, shifting the mood from wow to whoa. One joker dropped the line of the day: “does anyone have a Mumbai-based SaaS orchestrator for my orchestrators?” — a dunk on the endless parade of tool orchestration startups. Meanwhile, a straight-to-the-point commenter asked the awkward question everyone wondered: What’s the pricing?
So the vibes? Excited about the capabilities, skeptical about the “open source” label, amused by the orchestrator-of-orchestrators meme, and impatient for clear pricing. Dexto’s got heat — now it needs clarity.
Key Points
- •Dexto is a toolkit and intelligence layer for building agentic, stateful AI agents with memory.
- •It provides a YAML-based framework, a runtime for orchestration and multimodal support, and interfaces including CLI, Web UI, APIs, and a TypeScript SDK.
- •Dexto supports 50+ LLMs (OpenAI, Anthropic, Google, Groq, local models) and integrates 30+ tools via MCP clients and servers.
- •It offers persistent sessions, human-in-the-loop approval policies, observability with OpenTelemetry, and pluggable storage (Redis, PostgreSQL, SQLite, S3).
- •Deployment is flexible across local, cloud, or hybrid environments, with quick start instructions and an auto-approve flag for faster prototyping.