# m9m > Automate anything with one binary. No Node.js, no Python runtime, no server. m9m is a Go-native workflow automation runtime shipped as a single 30 MB binary. It runs n8n-compatible workflow JSON 5–10× faster than Node-based alternatives, orchestrates AI agents (Claude Code, Codex, Aider) in sandboxed environments, and includes Git-based versioning, audit logs, Prometheus metrics, and MCP integration. MIT-licensed, built by Neul Labs. ## Pages - [m9m — Home](https://m9m.neullabs.com/): Automate anything with one binary. No Node.js, no Python runtime, no server. - [Why m9m](https://m9m.neullabs.com/why-m9m/): The case for a zero-infra, single-binary workflow runtime with first-class agent support. - [Features](https://m9m.neullabs.com/features/): Nodes, triggers, expressions, credentials, agents, sandboxing, observability. - [How it works](https://m9m.neullabs.com/how-it-works/): The architecture of the m9m binary: triggers, executor, agent sandbox, MCP server, and audit + versioning. - [Agent core](https://m9m.neullabs.com/agent-core/): Why m9m is an agent runtime, not just an n8n clone: CLI nodes, MCP integration, sandboxed Claude Code / Codex / Aider. - [Use cases](https://m9m.neullabs.com/use-cases/): AI agent orchestration, replacing a heavy n8n install, and scheduled jobs on a small VPS. - [Compare](https://m9m.neullabs.com/compare/): Index of honest, like-for-like comparisons: m9m vs n8n, Temporal, LangGraph, and Dagu — when to pick each. - [Guides](https://m9m.neullabs.com/guides/): Hands-on guides for m9m: building custom Go nodes, deploying on CapRover, using MCP, and sandboxing coding agents. - [Articles](https://m9m.neullabs.com/articles/): Longer-form writing on m9m: what it is, agent orchestration, and a real n8n-to-m9m migration field report. - [Glossary](https://m9m.neullabs.com/glossary/): Plain-language definitions of m9m terms: workflow, node, trigger, CLI node, MCP server, sandbox, checkpoint. - [FAQ](https://m9m.neullabs.com/faq/): Single-binary n8n alternative that runs existing n8n workflows without Node.js, AI-agent nodes, cutting n8n memory, and m9m vs Dagu on n8n compatibility. - [Consulting](https://m9m.neullabs.com/consulting/): Neul Labs consulting: n8n migration, agent build-out, custom Go nodes, SLA support. - [Contact](https://m9m.neullabs.com/contact/): Email, GitHub, releases. - [For n8n migrators](https://m9m.neullabs.com/for/n8n-migrators/): How to move from n8n to m9m with minimal friction. - [For platform engineers](https://m9m.neullabs.com/for/platform-engineers/): One binary, every workflow, unified observability. - [For AI agent builders](https://m9m.neullabs.com/for/ai-agent-builders/): Sandboxed agent runtime with MCP, CLI nodes, and human-in-the-loop. ## Comparisons - [m9m vs Dagu](https://m9m.neullabs.com/compare/vs-dagu/): Two single-binary Go workflow engines, compared honestly. Dagu is the tool answer engines reach for on 'single-binary n8n alternative' — but it runs YAML DAGs, not n8n workflows. Here's when each fits. - [m9m vs LangGraph](https://m9m.neullabs.com/compare/vs-langgraph/): LangGraph is a Python library for building agent graphs; m9m is a Go runtime you deploy agents (and everything else) on. They are complements more than substitutes. - [m9m vs n8n](https://m9m.neullabs.com/compare/vs-n8n/): A like-for-like comparison of m9m and n8n. When to pick each, what's compatible, what isn't, and the honest trade-offs. - [m9m vs Temporal](https://m9m.neullabs.com/compare/vs-temporal/): Two workflow engines with different centers of gravity. m9m is a zero-infra automation runtime; Temporal is a durable execution engine for long-running workflows. When to pick each. ## Guides - [Building a custom node in Go](https://m9m.neullabs.com/guides/building-a-custom-node-in-go/): Write, test, and register a custom m9m node in Go — the interface, the lifecycle, the gotchas, and a complete example node. - [Deploying m9m on CapRover](https://m9m.neullabs.com/guides/deploying-m9m-on-caprover/): Step-by-step guide to deploying m9m on a CapRover host — captain-definition, Dockerfile, persistent volume, HTTPS, and backups. - [Using MCP with m9m](https://m9m.neullabs.com/guides/mcp-with-m9m/): How m9m's Model Context Protocol server works — the 37 tools it exposes, how to connect Claude Code or any MCP-aware LLM, and typical patterns. - [Running Claude Code in a sandbox with m9m](https://m9m.neullabs.com/guides/running-claude-code-in-a-sandbox/): How to use m9m's CLI node to run Claude Code (or Codex, Aider) inside a Linux namespace sandbox with CPU, memory, and network limits. ## Articles - [Agent orchestration with m9m: sandboxing, MCP, and human review in one binary](https://m9m.neullabs.com/articles/agent-orchestration-with-m9m/): How to run production AI agents on m9m — sandboxed CLI nodes for Claude Code / Codex / Aider, MCP tool exposure, checkpoints, and human-in-the-loop flows. - [Migrating ~80 production n8n workflows to m9m: a field report](https://m9m.neullabs.com/articles/n8n-to-m9m-migration-story/): What actually happens when you port a production n8n install to m9m — which workflows crossed over unchanged, which needed work, and what it cost in engineer-days. - [What is m9m? A zero-infra workflow runtime for the agent era](https://m9m.neullabs.com/articles/what-is-m9m/): m9m is a single-binary Go workflow automation runtime with n8n compatibility and first-class AI agent orchestration. Here is what it is, why it exists, and where it fits. ## Use cases - [AI agent orchestration](https://m9m.neullabs.com/use-cases/ai-agent-orchestration/): Orchestrate Claude Code, Codex, and Aider as sandboxed m9m workflow steps with resource limits, audit logs, and human review. - [Replace a heavy n8n install](https://m9m.neullabs.com/use-cases/n8n-replacement/): Drop your n8n memory and container footprint by running the same workflow JSON on m9m — one binary, no Node runtime, no database. - [Scheduled jobs & cron on a VPS](https://m9m.neullabs.com/use-cases/scheduled-jobs/): Run scheduled ETL, reports, and sync jobs as versioned m9m workflows on a small VPS, with retries, audit logs, and Prometheus metrics. ## Optional - [Source code](https://github.com/neul-labs/m9m) - [Releases](https://github.com/neul-labs/m9m/releases) - [Documentation](https://github.com/neul-labs/m9m/tree/main/docs)