Open foundations for AI in networking and security.
WireSage is an open-source community building the shared foundations for AI in networking and security: MCP servers, agent skills, language tools and plugins that let AI agents work with real networks. It starts with MikroTik RouterOS and aims at any platform, in public, under the MIT license.
WireSage open-source projects
What is WireSage?
Network and security tools are documented in RFCs, vendor manuals and decades of CLI habit. Language models read all of it fluently, but reading is not operating, and neither is defending.
Our hypothesis is that the gap is an interface problem, not a model problem. Give models small, typed, well-described tools, and the knowledge to use them, and they become useful colleagues instead of risky ones. WireSage builds those pieces in public: servers that act, skills that teach, and language tools that check.
It starts with MikroTik RouterOS. The pattern is not vendor-specific, and neither is the goal: the same foundations for any network or security platform.
- Typed tools, not raw shells
- Read-only on request
- Measured, not guessed
- MIT-licensed, built in public
- 01skills → agentloads routeros-bridging-switching
- 02agent → mcptools/call mikrotik_create_vlan_interface
- 03mcp → routerssh · /interface vlan add name=vlan200
- 04router → mcpok
- 05mcp → agentvlan200 created on ether1 · tag 200
Bring a tool, a lab topology or a bug report.
Maintained by volunteers; every contribution size counts. Start with a good first issue, or open an issue about an integration we haven’t touched yet.
$ gh search issues --owner wiresage --label "good first issue" --state open