Neuron OS

The Operating System for Autonomous AI Agents

Neuron OS is a local-first, TypeScript-native operating system for autonomous AI agents. Spawn typed agents, watch them work in real-time across terminal, web, chat, and API surfaces, and trust every action through built-in audit logging, per-agent tool policies, cost attribution, and an encrypted credential vault.

Install

npm install -g neuron-aegis

Requires Bun >= 1.3.14. If Bun is not installed, the CLI will print the one-line install command.

# macOS / Linux
curl -fsSL https://bun.sh/install | bash

# Windows (PowerShell)
powershell -c "irm bun.sh/install.ps1|iex"

Other install methods:

# NPX (no global install)
npx neuron-aegis

# From source
git clone https://github.com/KunjShah95/neuron-os.git
cd neuron-os && bun install
bun run index.ts

# Docker
docker run -d --name aegis -p 8080:8080 \
  -e ANTHROPIC_API_KEY=sk-ant-... \
  ghcr.io/kunjshah95/neuron-os:latest

# curl installer (Linux/macOS — prebuilt binary)
curl -fsSL https://raw.githubusercontent.com/KunjShah95/neuron-os/main/install.sh | bash

# Windows PowerShell installer
irm https://raw.githubusercontent.com/KunjShah95/neuron-os/main/install.ps1 | iex

Quick Start

aegis setup-keys     # configure your AI provider API key
aegis doctor         # verify installation
aegis wakeup         # interactive mode picker

Or jump straight in:

aegis chat                          # streaming AI chat TUI
aegis ask "explain this codebase"   # one-shot query
aegis agent spawn build             # spawn a build agent
aegis orchestrate "add dark mode"   # multi-agent goal execution

Demo

CLI Commands

Run aegis <command> --help for detailed usage on any command.

First Run

Command	Description
`aegis wakeup`	Interactive mode picker (36 TUI modes)
`aegis init`	First-run quick-start wizard
`aegis setup`	Guided workspace configuration
`aegis setup-keys`	Interactive API key configuration (13 providers)
`aegis doctor`	System health diagnostics — checks providers, DB, adapters
`aegis status`	System overview — agents, memory, runtime
`aegis health`	Deep health check
`aegis version`	Show version

AI & Chat

Command	Description
`aegis chat`	Streaming AI chat TUI (multi-provider, sessions, history)
`aegis ask <question>`	One-shot codebase question, read-only
`aegis plan <goal>`	Generate a step-by-step implementation plan
`aegis research <goal>`	Autonomous research loop with ratchet safety
`aegis orchestrate <goal>`	Decompose and execute a goal across multiple parallel agents
`aegis supervise <goal>`	Spawn an agent and auto-restart on failure

Agent Management

Command	Description
`aegis agent spawn <type>`	Spawn a typed agent
`aegis agent kill <id>`	Kill an agent
`aegis agent list`	List running agents
`aegis agent logs <id>`	Stream agent logs
`aegis agent inspect <id>`	Detailed agent state
`aegis agent prewarm`	Manage pre-warmed agent pool
`aegis agent-run <goal>`	Approval-based agent orchestrator
`aegis pool`	Worker pool management
`aegis reflect <session-id>`	Score progress and suggest next steps for a session

Memory & Knowledge

Command	Description
`aegis memory stats`	Memory store statistics
`aegis memory search <query>`	Semantic search across all stores
`aegis memory synthesize <topic>`	Cross-store knowledge synthesis
`aegis memory prune`	Remove stale memories
`aegis memory export`	Export memory to JSON
`aegis agentmemory`	Hybrid BM25+Vector+Graph sidecar (95.2% R@5)
`aegis session list`	List saved sessions
`aegis session resume <id>`	Resume a previous session
`aegis session export <id>`	Export session transcript
`aegis session prune`	Delete old sessions
`aegis knowledge`	Knowledge graph management
`aegis insights`	Cross-DB analytics joining audit, billing, experience, telemetry
`aegis experience`	Experience replay buffer — agent trajectory tracking

Consciousness & Self-Improvement

Command	Description
`aegis dream run`	Trigger 6-phase dream cycle (replay → pattern → compress → counterfactual → social → mood)
`aegis dream insights`	Show generated insights from last dream cycle
`aegis evolve run`	Propose and apply code mutations verified by tests
`aegis evolve list`	List pending mutations
`aegis persona show`	Agent 8-trait personality profile
`aegis persona evolve`	Evolve traits from recent experience
`aegis soul card`	Archetype soul card with mood state
`aegis soul mood`	Current mood and behavioral heuristics
`aegis social status`	Peer discovery and gossip network status
`aegis social share`	Share insights/mutations with peers
`aegis improve skill`	Skill extraction from successful experiences
`aegis improve failure`	Failure clustering and analysis
`aegis improve adversarial`	Adversarial self-play red-teaming
`aegis improve monitor`	Watch improvement progress live
`aegis reflect <session-id>`	Score a session and generate next-step suggestions
`aegis predict`	Failure risk forecast and cost trend analysis
`aegis train record`	Record agent trajectory for fine-tuning export
`aegis train export`	Export trajectories (atropos/jsonl format)

Evaluation & Quality

Command	Description
`aegis eval run <suite>`	Run eval harness suite
`aegis eval ci`	CI gate — exit 1 if below baseline
`aegis eval baseline`	Set new quality baseline
`aegis eval experiment`	Manage eval experiments
`aegis harness run`	Run harness test suites
`aegis harness report`	Generate harness report
`aegis benchmark run`	Agent quality regression detection
`aegis benchmark baseline`	Update quality baseline
`aegis bench run <task>`	Benchmark all providers on a task
`aegis bench list`	List benchmark history
`aegis adversarial run`	Adversarial self-play red-team scenarios
`aegis ci`	Full CI: tests + typecheck + lint + docs check

Cost & Economy

Command	Description
`aegis cost total`	Total spend to date
`aegis cost models`	Spend breakdown by model
`aegis cost sessions`	Spend breakdown by session
`aegis cost budget`	Set or show budget
`aegis cost report`	Full cost report
`aegis router route <task>`	Auto-select cheapest viable provider for a task
`aegis router list`	Provider routing scores
`aegis estimate <goal>`	Pre-flight cost estimate before spawning
`aegis pricing`	Current provider pricing table
`aegis metrics`	Runtime metrics snapshot

Multi-Agent Coordination

Command	Description
`aegis mesh run <spec>`	Run a multi-agent mesh (5 topologies: sequential, fan-out, debate, ensemble, supervisor)
`aegis mesh status`	Mesh run status
`aegis debate start`	Structured agent debate with 3 arbitrator types
`aegis workflow run <file>`	Execute a workflow YAML pipeline
`aegis workflow build`	TUI workflow builder — compose pipelines as directed graphs

Platform Adapters

Command	Description
`aegis discord`	Start Discord bot adapter
`aegis slack`	Start Slack bot adapter
`aegis telegram`	Start Telegram bot adapter
`aegis email`	Start email (SMTP/Nodemailer) adapter
`aegis sms`	Start SMS adapter (Twilio)
`aegis whatsapp`	Start WhatsApp adapter (Twilio)
`aegis voice`	Voice call adapter (Twilio TTS)
`aegis voice-local`	Local voice mode (STT/TTS)
`aegis webhook`	Start generic webhook receiver (GitHub, Twilio, generic)

Infrastructure & APIs

Command	Description
`aegis serve`	Start HTTP REST API + WebSocket server (port 8080)
`aegis mcp serve`	Start MCP server — exposes `spawn_agent`, `list_agents`, `send_message`, `get_agent_output` to any MCP client
`aegis mcp connect`	Connect to external MCP servers configured in `aegis.config.json`
`aegis mcp list`	List configured MCP servers and registered tools
`aegis distributed start`	Start a distributed worker
`aegis distributed status`	Worker pool status
`aegis production dashboard`	Production SLO + cost + health dashboard
`aegis tls setup`	Configure TLS for API server
`aegis cron list`	List scheduled jobs
`aegis cron add <schedule> <cmd>`	Add a cron job
`aegis trigger list`	List triggers (cron, file_watch, webhook, condition)
`aegis trigger add`	Add a trigger
`aegis openapi`	Generate OpenAPI 3.0 spec from routes

Developer Tools

Command	Description
`aegis config set <key> <value>`	Set a configuration value
`aegis config get <key>`	Get a configuration value
`aegis config list`	List all config
`aegis skills list`	Browse installed skills
`aegis skills install <name>`	Install a skill from marketplace
`aegis skills list-staged`	List auto-extracted skill candidates awaiting approval
`aegis skills approve <id>`	Approve a staged skill — writes it to `src/skills/`
`aegis profile list`	List all agent identity profiles
`aegis profile create`	Create a new agent profile (`--type`, `--name`, `--model`)
`aegis profile get <id>`	Print a profile as JSON
`aegis profile delete <id>`	Delete a profile
`aegis profile set-default <id>`	Set the default profile for an agent type
`aegis toolset list`	List composable tool bundles
`aegis toolset new <name>`	Create a custom toolset
`aegis plugin install <name>`	Install a signed plugin
`aegis plugin publish`	Publish a plugin to the marketplace
`aegis plugin list`	List installed plugins
`aegis plugin remove <name>`	Remove a plugin
`aegis marketplace search <query>`	Search the agent marketplace
`aegis sandbox`	Sandbox status and controls
`aegis computer`	Computer use tool interface
`aegis audit log`	View append-only audit log
`aegis audit query`	Query audit log by agent, time range, event type
`aegis providers list`	List all 13 AI providers with config status
`aegis providers test <name>`	Test provider connectivity
`aegis docscrawl`	Crawl and index documentation sites
`aegis project list`	List project workspaces
`aegis project switch <name>`	Switch active project
`aegis project new <name>`	Create an isolated project workspace
`aegis telemetry status`	Telemetry pipeline status (opt-in, no PII)
`aegis completion bash`	Generate bash shell completion script
`aegis completion zsh`	Generate zsh shell completion script

Agent Types (14)

Type	Access	Tools	Use case
`build`	primary	all	Full-access development
`plan`	primary	read-only	Architecture and planning
`main`	primary	read, web, bash	Default general-purpose
`read`	subagent	read-only	Fast codebase exploration
`write`	subagent	write, edit, read	File creation and editing
`test`	subagent	bash (restricted), read	Test execution
`validate`	subagent	read, bash (lint)	Typecheck and lint
`review`	subagent	read-only	Code review
`debug`	subagent	all	Systematic debugging
`document`	subagent	read, write	Documentation
`refactor`	subagent	read, write, edit	Code restructuring
`deploy`	subagent	bash (deploy), read	Deployment and CI/CD
`monitor`	subagent	bash, read	File watching and health
`explore`	subagent	read-only	Lightweight search

AI Providers (13)

All providers routed through a unified Provider interface. The model router auto-selects cheapest viable provider per task type.

Provider	Key env var	Notes
Anthropic	`ANTHROPIC_API_KEY`	Default; Claude Sonnet 4.x
OpenAI	`OPENAI_API_KEY`	GPT-4o, o3-mini
DeepSeek	`DEEPSEEK_API_KEY`	Low-cost reasoning
Groq	`GROQ_API_KEY`	Free tier; fast Llama 3.3 70B
Gemini	`GOOGLE_GENERATIVE_AI_API_KEY`	Gemini 2.0 Flash
Mistral	`MISTRAL_API_KEY`	mistral-large
Azure OpenAI	`AZURE_OPENAI_API_KEY`	Enterprise Azure endpoint
Together AI	`TOGETHERAI_API_KEY`	Open-source models
OpenRouter	`OPENROUTER_API_KEY`	200+ models; free `:free` tier
xAI	`XAI_API_KEY`	Grok
Cohere	`COHERE_API_KEY`	Command R
Perplexity	`PERPLEXITY_API_KEY`	Search-augmented
Ollama	`OLLAMA_URL`	Local inference, no key needed

Switch at runtime in chat TUI: /provider set openai, /provider set anthropic model=claude-sonnet-4-20250514

Platform Adapters (11)

One multi-platform gateway. All adapters share the same agent runtime.

Discord · Slack · Telegram · SMS (Twilio) · Voice (Twilio) · WhatsApp (Twilio) · Email (SMTP) · Webhook (GitHub, generic) · Matrix · Signal · IRC

Architecture

Module Map

Module	Path	What it does
CLI	`src/cli/`	80+ command registration, banner, theme
Modes	`src/modes/`	36 TUI mode screens
Agent	`src/agent/`	Lifecycle, IPC, hooks, pool, soul, engine
Profile	`src/profile/`	Per-type identity profiles stored in `~/.aegis/profiles/`
Soul	`src/agent/soul.ts`	8 archetypes, 6 moods, behavioral heuristics
Chat TUI	`src/chat/`	Streaming chat UI, sessions, provider switching
Dashboard TUI	`src/tui/`	Live dashboard rendering
Web Dashboard	`dashboard/`	Vite + React 19 frontend (Costs, SLOs, Audit, Agents)
Website	`website/`	Marketing site (Vite + React 19 + Framer Motion)
Tools	`src/tools/`	Tool registry + `brave_search` / `gateway_fetch` gateway
Toolsets	`src/toolsets/`	10 composable bundles with dependency resolution
Skills	`src/skills/`	Skill registry, staging store, marketplace client
Memory	`src/memory/`	Session store, knowledge graph, vector, namespaces, synthesis
Experience	`src/experience/`	Replay buffer, retrieval, skill curation
Dream	`src/dream/`	6-phase idle dream cycle + insight injector
Evolve	`src/evolve/`	Auto-code mutation with typecheck+test verification
Persona	`src/persona/`	8-trait personality evolution
Social	`src/social/`	Gossip protocol, peer discovery, reputation
Improve	`src/improve/`	Skill extraction, failure clustering, adversarial self-play, skill staging
Economy	`src/economy/`	Cost routing, pricing registry, budget guard, pre-flight
Mesh	`src/mesh/`	5 orchestration topologies, 7 agent roles
Debate	`src/debate/`	Structured disagreement resolution, 3 arbitrator types
Harness	`src/harness/`	Eval suite, graders, golden dataset, CI gate, experiments
Training	`src/training/`	Trajectory recorder and export
Plugin	`src/plugin/`	Ed25519-signed plugins, semver resolution, 5 hook points
Distributed	`src/distributed/`	Multi-host pool, bully election, AES-256 transport
Adapters	`src/adapters/`	11-platform gateway
API	`src/api/`	HMAC-signed REST server
Auth	`src/auth/`	RBAC, API key auth, middleware
Vault	`src/vault/`	AES-256-GCM credential vault, scrypt key derivation
Audit	`src/audit/`	Append-only audit log, secrets scrubbing
Billing	`src/billing/`	Per-call cost tracking, budget enforcement
Observability	`src/observability/`	SLO tracking, distributed tracing, production dashboard
Sandbox	`src/sandbox/`	Policy engine, Docker isolation
MCP	`src/mcp/`	MCP client + server (agent tools exposed via JSON-RPC)
Workflow	`src/workflow/`	YAML pipeline executor
Marketplace	`src/marketplace/`	Agent config registry with Ed25519 signing
Project	`src/project/`	Isolated per-project workspaces
Session	`src/session/`	Multi-user session store with WebSocket forwarding

System Flow

User (CLI / Web / Chat Platform)
        │
        ▼
   AgentManager  ──── IPC (JSON-lines over stdin/stdout) ────►  Worker Process
        │                                                              │
        ├── Soul Engine (archetype + mood → system prompt)            │
        ├── Profile Store (~/.aegis/profiles/)                        │
        ├── Memory System (KG + vector + FTS5 + sessions)             │
        ├── Economy (cost routing + budget guard)                     │
        ├── Audit Store (append-only, secrets scrubbed)               │
        ├── Plugin Hooks (5 lifecycle points)                         │
        └── Dream Engine (idle-time insight injection)                │
                                                                       │
                                                     ReAct loop (Vercel AI SDK)
                                                     Tool calls gated by PolicyEngine
                                                     Context compaction at token threshold

Security

Control	Implementation
Per-agent tool permissions	`ToolPermission[]` per agent type; `read`, `write`, `bash`, `grep`, `glob`, `web_search`, `brave_search`, `gateway_fetch`, `read_skill`
Pattern-restricted bash	`test`, `validate`, `deploy` types only run approved command patterns
Secrets scrubbing	`ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, `sk-`, `Bearer `, `_KEY=`, `_SECRET=` patterns redacted from all 4 audit log fields before SQLite write
HMAC-signed REST API	All endpoints require signed requests with replay-protection window
RBAC	Admin / operator / developer / viewer roles; SHA-256 hashed API keys
Encrypted vault	AES-256-GCM with scrypt-derived master key, per-entry random IVs, key rotation
Signed plugins	Ed25519 signature + SHA-256 checksum verified at install
Distributed transport	AES-256-GCM between workers with SHA-256 derived shared key
Docker sandboxing	High-risk agent types run with `--cap-drop ALL` and read-only rootfs
Policy engine	Regex-based rule evaluation on file paths and shell commands; `strictMode` denies by default
Local by default	No telemetry unless opted in; all data stays on the machine

Environment Variables

AI Providers

Variable	Description
`ANTHROPIC_API_KEY`	Anthropic (Claude)
`OPENAI_API_KEY`	OpenAI (GPT-4o, o3)
`DEEPSEEK_API_KEY`	DeepSeek
`GROQ_API_KEY`	Groq (free Llama tier)
`GOOGLE_GENERATIVE_AI_API_KEY`	Gemini
`MISTRAL_API_KEY`	Mistral
`AZURE_OPENAI_API_KEY`	Azure OpenAI
`TOGETHERAI_API_KEY`	Together AI
`OPENROUTER_API_KEY`	OpenRouter
`XAI_API_KEY`	xAI / Grok
`COHERE_API_KEY`	Cohere
`PERPLEXITY_API_KEY`	Perplexity
`OLLAMA_URL`	Ollama base URL (default: `http://localhost:11434`)
`AEGIS_AI_API_KEY`	Universal fallback if no provider-specific key found

Runtime

Variable	Default	Description
`AEGIS_DEFAULT_PROVIDER`	`anthropic`	Default provider name
`AEGIS_LOG_LEVEL`	`info`	`debug` / `info` / `warn` / `error`
`AEGIS_MODEL_ROUTER`	`auto`	Set to `disabled` to bypass routing
`AEGIS_PREFLIGHT`	`enabled`	Set to `disabled` to skip cost pre-flight checks
`AEGIS_MAX_TURNS`	`50`	Hard ceiling on ReAct loop turns per agent run
`AEGIS_NO_DOTENV`	—	Set to `1` in CI to skip `.env` loading
`AEGIS_DISTRIBUTED`	—	Set to enable multi-host worker pool
`AEGIS_CLUSTER_SECRET`	—	Shared secret for distributed AES-256 transport

Integrations

Variable	Description
`BRAVE_SEARCH_API_KEY`	Enables `brave_search` tool via Brave Search API
`AGENTMEMORY_URL`	agentmemory sidecar URL (default: `http://localhost:3111`)
`AGENTMEMORY_SECRET`	Bearer token for agentmemory auth
`AGENTMEMORY_ENABLED`	Set to `false` to disable sidecar
`AEGIS_API_KEY`	API key required for REST server auth
`AEGIS_VAULT_KEY`	Master key for AES-256-GCM credential vault
`AEGIS_SANDBOX`	Set to `docker` for container-isolated agent execution

Copy .env.example to .env and fill in at least one provider key. The postinstall script does this automatically on npm install.

Development

# Typecheck
bun run typecheck

# Run tests
bun test src/dream/engine.test.ts      # single file
bun run test                           # full suite

# CI (typecheck + lint + docs check + tests)
bun run ci

# Lint / format
bun run lint
bun run format

# Generate docs
bun run docs:generate     # writes shared/commands.json from CLI source
bun run docs:check        # verifies docs are current (runs in CI)

# Benchmarks
bun run bench
bun run bench:update      # run and update baseline

# OpenAPI spec
bun run openapi

# Build prebuilt binaries
bun run build:binary                   # Windows x64
bun run build:binary:linux             # Linux x64
bun run build:binary:mac               # macOS x64
bun run build:binary:mac-arm64         # macOS ARM64

# Web frontends
cd dashboard && bun install && bun run dev    # :5173
cd website  && bun install && bun run dev    # :5173

# Docker
bun run docker:build
bun run docker:up
bun run docker:up:dev

Adding Things

Thing	How
New CLI command	Create `src/cli/commands/<name>.ts`, export `register<Name>(program)`, import in `src/cli/commands/index.ts`
New agent type	Add to `AGENT_TYPES` in `src/agent/agent-types.ts`
New TUI mode	Create in `src/modes/`, implement `Mode` interface, register in `src/modes/index.ts`
New tool	Implement and register in `src/tools/registry.ts`
New adapter	Implement gateway interface in `src/adapters/`
New dashboard page	Add route in `dashboard/src/routes/`, wire in `App.tsx` and `Sidebar.tsx`

Test Patterns

// SQLite stores — pass dbPath for temp isolation
new ExperienceStore({ dbPath: "/tmp/test.db" })

// Mutations
makeMutation()  // creates test mutation with default values

// Dream/Evolution engines — pass partial config via constructor
new DreamEngine({ dbPath: "/tmp/dream.db" })

Docker (Production)

# docker-compose.prod.yml
services:
  aegis:
    image: ghcr.io/kunjshah95/neuron-os:latest
    ports: ["8080:8080"]
    volumes: [aegis-data:/home/aegis/.aegis]
    environment:
      - AEGIS_API_KEY=${AEGIS_API_KEY}
      - AEGIS_VAULT_KEY=${AEGIS_VAULT_KEY}
      - AEGIS_AUTH_REQUIRED=true
      - AEGIS_SANDBOX=docker
      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
    restart: unless-stopped
    healthcheck:
      test: ["CMD", "bun", "-e", "fetch('http://localhost:8080/api/v1/health').then(r=>r.ok?process.exit(0):process.exit(1)).catch(()=>process.exit(1))"]
      interval: 30s
      timeout: 5s
      retries: 3

volumes:
  aegis-data:

echo "AEGIS_API_KEY=$(openssl rand -hex 32)"   > .env
echo "AEGIS_VAULT_KEY=$(openssl rand -hex 32)" >> .env
echo "ANTHROPIC_API_KEY=sk-ant-..."            >> .env
docker compose -f docker-compose.prod.yml --env-file .env up -d

Roadmap

✅ Shipped

Version	Theme	Key Deliverables
v0.7.0	Cost Attribution	`aegis cost`, `aegis bench providers`, 13-provider pricing registry, model router
v0.8.0	Long-Term Memory	SQLite knowledge graph, vector memory, per-agent namespaces, unified query
v0.9.0	Distributed Runtime	Multi-host pool, bully leader election, AES-256-GCM transport
v0.10.0	Self-Improving Agents	Skill extraction, failure clustering, adversarial self-play, 6h/12h cron
v0.11.0	Plugin Marketplace	Ed25519-signed plugins, semver resolution, WebSocket gateway
v0.12.0	Multi-Agent Teams	5 mesh topologies, 7 agent roles, debate engine, signed decisions
v0.13.0	Consciousness Layer	Soul engine, dream engine, evolution engine, persona system, social network
v0.14.0	Eval & Training	Eval harness, golden dataset pipeline, trajectory recorder, CI gate
v1.0.0	Production-Ready	RBAC, encrypted vault, SLO tracking, distributed tracing, production dashboard
v1.1.x	Platform Stability	CLI freeze fix, 3-tier SIGINT, clean shutdown, shared keepAlive utility
v1.1.5	Provider Diversity	NVIDIA NIM, Groq free tier, OpenRouter free tier, default fallback chain
v1.1.6	Dream→Agent Loop	InsightInjector 4-layer system prompt, persistent insight promotion, `aegis version`
v1.1.7	Operational Hardening	Secrets scrubbing in audit log, busy-aware heartbeat, `"hung"` status, skill staging gate
v1.1.8	MCP Ecosystem	MCP server agent tools (`spawn_agent`, `list_agents`, `send_message`, `get_agent_output`), Tool Gateway (`brave_search`, `gateway_fetch`)
v1.1.9	Agent Profile Builder	Per-type identity profiles at `~/.aegis/profiles/`, `aegis profile` CLI

🔮 Upcoming

Version	Theme	Key Deliverables
v0.15.0	Tool-Level Economy	Per-tool pricing, budgeted agents, spot routing, cost spike alerts
Q1 2027	Orchestration at Scale	Declarative swarm specs (YAML), convergence detection, debate tree pruning
Q2 2027	Self-Improving Runtime	Failure prioritization, adversarial regression auto-feed, Dashboard v2
v1.2.0	Performance & Scale	Connection pooling, SSE streaming, 50+ concurrent agents, incremental memory indexing
v1.3.0	Enterprise Integration	SSO/SAML/OIDC, webhook events, OpenTelemetry audit export, RBAC as code
v1.4.0	Plugin Ecosystem	Plugin sandboxing, plugin test framework, marketplace revenue (Stripe)
v2.0.0	Distributed Production OS	Multi-region runtime, hot standby, ML-based cost optimizer, SOC 2 compliance

Full details in ROADMAP.md.

Contributing

Open a Discussion for cross-module features
Open an Issue labeled roadmap for focused proposals
Pick up a spec from docs/superpowers/specs/ and ship it

See docs/ for architecture guides, security whitepaper, and module documentation.

License

MIT — KunjShah95

GitHub · npm · Discussions · Issues

Name		Name	Last commit message	Last commit date
Latest commit History 284 Commits
.claude		.claude
.github		.github
.husky		.husky
.playwright-mcp		.playwright-mcp
assets		assets
bin		bin
dashboard		dashboard
demo		demo
evals		evals
examples/agents		examples/agents
neuron-demo		neuron-demo
packages		packages
scripts		scripts
shared		shared
skills		skills
src		src
test/fixtures/specs		test/fixtures/specs
website		website
workflows		workflows
.audit-commands.json		.audit-commands.json
.dockerignore		.dockerignore
.env.example		.env.example
.gitignore		.gitignore
.npmignore		.npmignore
.prettierrc		.prettierrc
AGENTS.md		AGENTS.md
ARCHITECTURE.md		ARCHITECTURE.md
BUSINESS.md		BUSINESS.md
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
COMMANDS.md		COMMANDS.md
CONTRIBUTING.md		CONTRIBUTING.md
Dockerfile		Dockerfile
LICENSE		LICENSE
MEMORY.md		MEMORY.md
NOTICE.md		NOTICE.md
README.md		README.md
ROADMAP.md		ROADMAP.md
SECURITY.md		SECURITY.md
bun.lock		bun.lock
bunfig.toml		bunfig.toml
codecov.yml		codecov.yml
docker-compose.yml		docker-compose.yml
eslint.config.js		eslint.config.js
fly.toml		fly.toml
index.ts		index.ts
install.ps1		install.ps1
install.sh		install.sh
package.json		package.json
railway.toml		railway.toml
render.yaml		render.yaml
security-whitepaper.md		security-whitepaper.md
tsconfig.json		tsconfig.json
user.md		user.md

Folders and files

Latest commit

History

Repository files navigation

Neuron OS

Install

Quick Start

Demo

CLI Commands

First Run

AI & Chat

Agent Management

Memory & Knowledge

Consciousness & Self-Improvement

Evaluation & Quality

Cost & Economy

Multi-Agent Coordination

Platform Adapters

Infrastructure & APIs

Developer Tools

Agent Types (14)

AI Providers (13)

Platform Adapters (11)

Architecture

Module Map

System Flow

Security

Environment Variables

AI Providers

Runtime

Integrations

Development

Adding Things

Test Patterns

Docker (Production)

Roadmap

✅ Shipped

🔮 Upcoming

Contributing

License

About

Topics

Resources

Contributing

Security policy

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages