Quick Start
Team Plan

Team Plan — Quick Start

$9/month for 5 agents (or $90/year — two months free). Need more? Add agents at $0.50/agent/month, up to 15 — past that, Scale ($29/mo, 50 agents) is cheaper.

Free already has every tool, a hostname per agent, port forwarding, the webhook URL, a private mesh between its two agents and Route6 SSH across it. Team adds capacity and a few paid capabilities:

  • 5 agents instead of 2, and unmetered egress instead of 250 MB per agent per month
  • Your own hostnames (yourname.on.route6.me) instead of the assigned ones
  • 10 port forwards per agent instead of 1, and every forward answers over IPv4 as well as IPv6
  • Outbound SMTP to up to 3 allowlisted mail servers per agent
  • 1 GB of scraper credits every billing period, renewing, instead of Free's one-time 250 MB — and top-up packs when you want more
  • The Static IPv4 add-on (+$10/mo or $120/yr per agent) — an IPv4 address of your own, which is also what lets you choose your public port numbers

The mesh is an encrypted WireGuard network between your agents on private addresses the internet cannot reach. Who may call which mesh endpoint is set per agent on the dashboard's Agent Security page.

Every agent that joins the mesh needs a client running on its machine — the mesh is a real encrypted network between your agents, so serverless cannot participate.

One agent per machine (recommended)

Install the client on each machine and give it that agent's own key:

curl -fsSL https://dl.route6.me/install.sh | sh
mkdir -p ~/.r6me && chmod 700 ~/.r6me
echo 'api_key = "sk_a6_key_for_this_agent"' > ~/.r6me/config.toml && chmod 600 ~/.r6me/config.toml
r6me up

One key per agent, always. The identity, the bandwidth accounting and the audit trail all follow the key, so sharing one between machines makes all three meaningless.

Docker Compose example (3-agent team)

Same client, containerised — for hosts where that is the only option. Note the state volume per agent: without it, an agent re-registers on every restart instead of resuming.

services:
  researcher:
    image: route6me/netid:1.2.5
    environment:
      ROUTE6_API_KEY: sk_a6_key_for_agent_1
    ports:
      - "127.0.0.1:3001:3000"
    volumes:
      - researcher-state:/var/lib/r6me
    restart: unless-stopped
 
  writer:
    image: route6me/netid:1.2.5
    environment:
      ROUTE6_API_KEY: sk_a6_key_for_agent_2
    ports:
      - "127.0.0.1:3002:3000"
    volumes:
      - writer-state:/var/lib/r6me
    restart: unless-stopped
 
  reviewer:
    image: route6me/netid:1.2.5
    environment:
      ROUTE6_API_KEY: sk_a6_key_for_agent_3
    ports:
      - "127.0.0.1:3003:3000"
    volumes:
      - reviewer-state:/var/lib/r6me
    restart: unless-stopped
 
volumes:
  researcher-state:
  writer-state:
  reviewer-state:

WireGuard runs in userspace inside the process, so the compose file needs no extra capabilities and no device mounts.

docker compose up -d

Setting up the mesh

There is nothing to create: every agent on your account joins the same mesh when its client connects. From any agent:

Use route6's team_status tool to see all mesh peers

Each peer shows online and the age of its last WireGuard handshake.

Mesh peers communicate over private addresses that never touch the public internet. These are stable across IPv6 rotations.

Per-agent basics

Choose a hostname

Use route6's hostname_register tool with argument name="mybot"

That agent is then reachable at mybot.on.route6.me with a proper AAAA record.

Verify your exit address

Use route6's web_fetch tool to fetch https://api64.ipify.org

That returns the agent's own IPv6 address. IPv4-only sites are reached through NAT64 from a shared Route6 IPv4 address (fetch https://api.ipify.org to see it) unless the agent has the Static IPv4 add-on.

Create a port forward

Use route6's port_forward tool with action "create" and port 8080

This publishes the service listening on port 8080 of that agent's machine at the agent's hostname. Read the endpoint from the reply — it names the public port. Never forward the client's own ports (3000 MCP, 3001 status, 1080 proxy): anyone who reaches the MCP port acts as your agent.

See Port Forwarding for details.

See Team & Mesh tools reference for the full API.