Design multi-agent workflows you can trust

Hey everyone! We’re a small team with an extensive background in AI, and naturally wanted to automate some of our own processes with agent workflows. We tried a bunch of tools, but none quite fit, so we built our own open-source framework: guilds Our approach is built around three principles: Boundaries with freedom: Give agents clear constraints while letting them handle real-world surprises. Observability: See what agents do, which tools they use, what assumptions they make, and how their state changes. Shareability: Fine-tuning complex workflows is hard, so we want teams to share working recipes with the community. If this sounds interesting, we’d love a ⭐ on github! The more people use it, the more useful recipes we can build together. If that doesn't resonate with you I'd love to know what your current workflow like.