MirrorNeuron Docs

Documentation for MirrorNeuron, including installation, architecture, CLI usage, blueprint development, security, and troubleshooting.

Project Scope

MirrorNeuron is an Elixir/BEAM runtime with Python tooling for running message-driven workflow graphs. The documentation covers:

Runtime architecture and clustering.
Local installation and first-run setup.
CLI, API, SDK, Web UI, blueprints, and skills.
Security considerations for running worker payloads.
Troubleshooting and contribution guidance.

Prerequisites

macOS, Linux, or WSL2.
Python 3.11+ for CLI, SDK, API, system tests, and Python-defined blueprints.
Docker for the default local Redis and core workflow.
Redis for runtime state.
OpenShell when running sandboxed workers.

Elixir/Erlang are required for core runtime development. Released-package installs use OTP tarballs instead of building the core from source.

Quick Start

Install MirrorNeuron from released packages:

curl -fsSL https://mirrorneuron.io/install.sh | bash

Start the runtime and run a sample blueprint:

mn runtime start
mn blueprint run message_routing_trace
mn blueprint monitor

Check jobs:

mn job list
mn node list

Architecture Summary

CLI / API / Web UI
      |
      v
BEAM runtime: jobs, agents, routing, leases, recovery, events
      |
      v
Execution runners: HostLocal, OpenShell, Python, shell, worker payloads
      |
      v
Redis durable state and optional Redis Sentinel HA

Read:

Documentation Index

Topic	Document
Repository/component guide	component-guide.md
Skill catalog	skill-catalog.md
Getting started	quickstart.md
Installation	installation.md
CLI reference	cli.md
Blueprints and skills	blueprints-and-skills.md
Blueprint standard	blueprint-standard.md
Docker and OpenShell for blueprints	docker_and_openshell_for_blueprints.md
Nomad-inspired runtime features	nomad-inspired-runtime.md
Services and health checks	services-and-health-checks.md
Model runtime	model-runtime.md
Resources and devices	resources-and-devices.md
Deployments	deployments.md
Schedules and events	schedules-and-events.md
Runtime architecture	runtime-architecture.md
Cluster architecture	cluster_architecture.md
Reliability and recovery	reliability.md
Security	security.md
Troubleshooting	troubleshooting.md
Contributing	contributing.md

Security Notes

MirrorNeuron can run local code, create sandboxes, call external services, connect to model providers, and pass selected environment variables to worker payloads.

Before running third-party bundles or exposing a cluster:

Keep runtime API and gRPC ports bound to trusted networks.
Review bundle manifest.json, payloads/, pass_env, and runner policies.
Use OpenShell policies for untrusted network access.
Do not pass broad secrets into workers.
Prefer Redis Sentinel HA over single Redis for multi-box deployments.
Treat incoming live messages, emails, Slack events, and model outputs as untrusted input.

Contributing

Keep docs concise and command examples copyable. When documenting behavior, link to the relevant component README or source file when practical.

License

See LICENSE.

MirrorNeuron Docs

On this page