Work in Progress
A Go library for orchestrating AI agents using directed graphs.
go-langgraph brings the power of LangGraph to Go, leveraging native concurrency (goroutines, channels, context.Context) and strong typing (generics) for production-grade AI agent orchestration.
- Directed graph execution — Define workflows as typed, directed graphs with statically validated edges.
- Fan-out / Fan-in — Native parallel execution of nodes with automatic state merging via reducers.
- Streaming — Channel-based event streaming for real-time observation.
- Checkpointing — Persist and resume execution at any step (in-memory, PostgreSQL, SQLite).
- Human-in-the-Loop — Interrupt execution for human review and resume with modified state.
- Retry & Circuit Breaker — Configurable retry policies with exponential backoff and circuit breaker per node.
- Multi-tenancy — Tenant isolation via context propagation.
- Observability — Structured events,
log/slogintegration, optional OpenTelemetry bridge. - Debug Studio — Embedded web UI for graph visualization, state inspection, and time-travel debugging.
- Zero dependencies — Core
graph/package depends only on the Go standard library.
- Go 1.24+
MIT