Give your AI agent its own place on the internet: a real public IPv6 address, a
*.on.route6.me hostname, port forwarding with a webhook URL, web fetch from its
own IP, and a private mesh with your other agents — 23 MCP tools, controlled by
the agent itself. Every tool is on every plan, including Free (no card).
In Claude Code:
/plugin marketplace add route6me/claude-plugin
/plugin install route6@route6
Then connect the MCP server. With the Route6 client running
(curl -fsSL https://dl.route6.me/install.sh | sh && r6me up, key from
https://route6.me):
claude mcp add --transport http route6 http://localhost:3000/mcpOr without a client (outbound tools only — no port forwards, webhooks or mesh):
claude mcp add --transport http route6 https://gw.route6.me/mcp \
--header "Authorization: Bearer $ROUTE6_API_KEY"The plugin ships the skill that teaches Claude when and how to use the tools — the webhook recipe, identity rotation and reputation checks, reaching teammates over the mesh, and team coordination patterns — plus a parameter reference for every tool, generated from the tool schemas.
| Area | Tools |
|---|---|
| Identity | identity (get / set_ipv6 / check_reputation), hostname_register |
| Inbound | port_forward (create / list / delete), port_forward_secure, port_forward_tls |
| Diagnostics | net (ping / traceroute / dns_resolve) |
| Web | web_fetch, web_search, web_browse, scrape |
smtp_allowlist |
|
| Team | team_status, team_ping, team_chat, team_whiteboard, team_capability, team_task, team_events, team_metrics, team_loop, team_project_task, team_roles |
| Plan | plan_upgrade |
Plans differ in capacity (agents, egress, scraper credits). A few capabilities need a paid plan — your own hostname, scraper top-ups, the Static IPv4 add-on, outbound SMTP — and the tool says so when called.
More: docs.route6.me · examples · route6.me/pricing
Plugin manifest and documentation: MIT. The Route6 service and client are proprietary.