goose is your on-machine AI agent, capable of automating complex development tasks from start to finish. More than just code suggestions, goose can build entire projects from scratch, write and execute code, debug failures, orchestrate workflows, and interact with external APIs autonomously. Whether you're prototyping an idea, refining existing code, or managing intricate engineering pipelines, goose adapts to your workflow and executes tasks with precision. Designed for maximum flexibility, goose works with any LLM and supports multi-model configuration to optimize performance and cost, seamlessly integrates with MCP servers, and is available as both a desktop app as well as CLI - making it the ultimate AI assistant for developers who want to move faster and focus on innovation.
External link for goose
goose reposted this
AI is great at analysing financial data...but only if you're willing to send your bank statements to the cloud. I wasn't. So I built a completely local workflow using goose, Docker Model Runner, and Gemma 4. With this, every statement, every prompt, and every generated report stays on my laptop. No API costs, no tokens to track, and no financial data leaving my machine. It's not fast, but it's 100% private. https://lnkd.in/d5dynwzz
Did you know you can run goose completely offline? Saiyam Pathak unplugged from the internet and ran goose on an NVIDIA DGX Spark by connecting to a local Ollama model over localhost. No cloud API calls, no data left the machine. https://lnkd.in/gZQQemn3
goose reposted this
Giving an agent every MCP tool a server exposes creates unnecessary risk. Andrew Zigler, AAIF Ambassador, shows how goose + agentgateway can federate MCP servers, scope tools per agent, route model and tool traffic through one gateway, and log every call with identity tracking. Build: https://bit.ly/4gBVKVs #MCP728
goose reposted this
✨Excited to be part of the first Agentic AI Foundation Ambassador AMA at MCP Developers Summit Seoul on August 13! 🇰🇷 Join Ambassadors Hoon Jo, Daniel Oh, Kevin Dubious, Junghwan Park, and me to ask us anything about MCP, goose, agentgateway, AGENTS.md, and contributing to the open agentic AI ecosystem Session: https://lnkd.in/eDzY4MaD
goose reposted this
The ability to choose and customize the surrounding layers of AI agents (tools, orchestration, guardrails) matters in production. That’s why Agentic AI can benefit from open infrastructure, open source strategy and #OSPO skills The open ecosystem at Agentic AI Foundation makes those layers easier to inspect and customize I share more in this The Linux Foundation article https://lnkd.in/eCymaN-y
If you give goose an MCP server, it gets every tool the server ships. Andrew Zigler shows how to put agentgateway in front of goose, federate MCP servers into one endpoint, scope down the tools, and log every model + tool call in one place. https://lnkd.in/gyZWJNJ7
goose reposted this
👩🔧 I’ve been exploring goose ’s built-in local inference together with Model Context Protocol servers 🧪 The results have been pretty fun! 📊 On the MCP side, I used a custom GitHub security-stats server based on CHAOSS metrics You can follow my learning in public in my latest article for the Agentic AI Foundation 🚀✨ https://lnkd.in/eFjv7QjP Thanks again AAIF for sharing this, and a special shoutout to CHAOSS ! 🙏 The Linux Foundation
Can a smaller local model outperform a larger one when working with MCP? In a hands-on experiment, Ana Jimenez Santamaria 🐧 tested goose’s local inference with two models, a custom MCP server, and a fully local agent stack. The result was a 4B model successfully drove multi-step tool calls where a 7B model struggled. The takeaway: agent performance is about orchestration, tools, and fit, not just model size. Read on: https://bit.ly/4vUV4PB
goose reposted this
Can a smaller local model outperform a larger one when working with MCP? In a hands-on experiment, Ana Jimenez Santamaria 🐧 tested goose’s local inference with two models, a custom MCP server, and a fully local agent stack. The result was a 4B model successfully drove multi-step tool calls where a 7B model struggled. The takeaway: agent performance is about orchestration, tools, and fit, not just model size. Read on: https://bit.ly/4vUV4PB
goose reposted this
hey divas~! 🐸✨ since ur zig is an Agentic AI Foundation Ambassador now, I'm gonna be serving up some more agentic engineering devlogs. and first one's up, hot n ready!! 🍞🍲 it's about agentgateway, which I only recently learned about, but have quickly fallen in love with! It tackles so many burning problems around MCP federation and model management, so I'm excited to show y'all a little taste of how I've been using it to chop down huge MCP servers into surgical tools for harnesses like goose to use with local models. Also inside is my /agentgateway skill that audits your tool calls and proposes smaller, virtualized MCP servers to remove unused stuff. Save tokens and reduce any blast radius, win win! And since loops are all the rage right now, yes, the skill is super loopable. I put this article on my own website (which recently underwent an agentic makeover -- it's OUR site now 😇) because my ambassador agent planned, screenshotted, and wrote a lot of it. We hope you enjoy & learn something! https://lnkd.in/gS6RQD6N #AAIFAmbassador #AAAmbassador
Goose is growing from an app into a platform. We're building GDK: the Goose Development Kit for building agentic apps with shared infra instead of reinventing the whole honking runtime. Let's see what's inside: orchestration, in-process local models, access to 50+ model providers, routing and dynamic model selection, context, tools, memory, remote execution ... (hold on, this goose needs a sip of water. whew!) ...automations and scheduling, skills, subagents, code mode, flexible agent loop, slash command infra, and more Companies and individual builders will be able to mix and match these building blocks to create purpose-built agentic apps for their own use cases. I'll honk to that!