agent-sdk-go
A Go AI agent SDK supporting durable execution via Temporal or zero-setup in-process mode, enabling rapid building of multi-model, tool-calling, and memory-equipped agents.
Tool overview
1. Adoption judgment: agent-sdk-go is in a very early stage, with only 30 GitHub stars and 1 fork, and no social media discussions, tutorials, or real-world case studies available. It should only be considered for experimental evaluation by developers experienced with Go and Temporal; production use is not recommended at this point. 2. What it does and barriers: The SDK provides native Go APIs to leverage models like OpenAI, Anthropic, Gemini and supports tool calling, MCP, A2A, RAG, conversation memory, streaming, sub-agent orchestration, and human-in-the-loop. It can run with Temporal for durable, crash-resilient execution or with zero setup in-process for fast prototyping. Required skills: Go proficiency; optional Temporal knowledge for durable mode. Cost: the project is open source under Apache-2.0; no official or community pricing/API fees were found (conservative inference: free to use, but external model API call fees apply separately). 3. Suitable/unsuitable audiences: Suitable for teams that want to build durable agents in Go, already use Temporal, or need workflow-driven AI. Also suitable for developers exploring Go-native agent frameworks.