MikroTik Router MCP
Manage MikroTik RouterOS networks through natural-language requests to an AI assistant.
- Source repo
- wiresage/mikrotik-mcp
- Stars
- ★ 316
- Last updated
- 1d ago
- License
- MIT
- Primary language
- Python
- FA score
- 54/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformClaude.ai (Partial support)
- Cost
- Free, no paid service needed
- Setup effort
- High · needs real infrastructure
- You'll need
- Typical use
- A network administrator asks an AI assistant to adjust VLANs on a MikroTik router.
- Not a fit if
- Teams that do not use MikroTik RouterOS devices
- Source review
- 54/100 · Major gaps 1 safety controls not found
What does this agent do, and when should you use it?
MikroTik MCP is an MCP server that bridges AI assistants and MikroTik RouterOS devices. It lets assistants carry out router-management operations from natural-language requests, including managing VLANs, configuring firewall rules, and handling DNS settings. The project says that, from version 0.14.0, users can manage multiple MikroTik devices through an inventory. It is intended for environments where an MCP client can connect to the server and reach the target routers. The README links to installation, testing, and Claude Desktop integration guides, but the supplied material does not include concrete deployment commands, runtime requirements, or authentication configuration.
An MCP client passes a natural-language request to MikroTik MCP, which interacts with MikroTik RouterOS devices to perform network-management operations. The README specifically names VLAN management, firewall-rule configuration, and DNS settings, but does not list exact callable tool names or a complete operation set. From version 0.14.0, the device inventory supports managing multiple devices. Results flow back to the AI assistant through the MCP interaction; the supplied material does not specify result formats or failure handling.
- A network administrator asks an AI assistant to adjust VLANs on a MikroTik router.
- A team responsible for router policy uses natural-language requests to configure MikroTik firewall rules.
- An administrator handles DNS settings on MikroTik devices through an assistant.
- An operator managing several MikroTik devices uses the inventory capability available from version 0.14.0.
- A user connecting MikroTik routers to Claude Desktop follows the integration guide linked by the project.
How do you install or deploy this agent?
The supplied material links to an installation guide and a Claude Desktop integration guide, but does not include installation steps, package or runtime requirements, a launch command, or device authentication instructions. Those details are needed for a reproducible installation; accurate commands cannot be given from the supplied material.
How do you use this agent?
The README says users can send natural-language requests through an AI assistant and links to Claude Desktop configuration and MCP Inspector interactive-testing guides. It does not provide the configuration, a tool-call example, or a first working request. To use it, configure an MCP client according to the relevant guide and provide connectivity and authentication for the target RouterOS device; the supplied material does not specify the required fields.
What are this agent's strengths and limitations?
- Connects AI assistant requests to MikroTik RouterOS devices through MCP.
- The README names concrete network operations: VLAN management, firewall-rule configuration, and DNS settings.
- Supports managing multiple MikroTik devices through an inventory from version 0.14.0.
- Links to integration guides for Claude Desktop and MCP Inspector.
- Its stated use depends on MikroTik RouterOS devices, so it does not fit teams managing other network-device platforms.
- The supplied material lacks installation commands, runtime versions, and launch configuration, making deployment effort difficult to assess.
- Device authentication, permission scope, and operation-confirmation behavior are not described in the supplied material.
- Callable tool names, the full feature set, and error handling are not documented in the supplied material.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| MikroTik Router MCP This agent | 54 · Major gaps | MCP serverFree | ★ 316 | 1d ago | Python | — |
| OptimNow Cloud FinOps Library | 65 · Some gaps | Agent plugin / skillFree | ★ 66 | today | Python | Claude Code |
| x-use | 55 · Major gaps | MCP serverFree | ★ 180 | 2d ago | Python | Codex · Claude Code · Claude.ai |
| AI Gateway Labs | 53 · Major gaps | CLIFree + model costs | ★ 1k | 1mo ago | Jupyter Notebook | — |
How does FollowAgents rate this agent?
Why each dimension lost points
The docs recommend restricted RouterOS accounts and describe a read-only tool catalogue, SSH-agent key restriction, HTTP IP/Host/Origin allowlists, and sensitive-data hiding; they also disclose automatic acceptance of unknown SSH host keys, unauthenticated HTTP transports, and container environment-variable exposure. This supports least-privilege and data-flow assessment. Deductions: write tools are enabled by default with no user-confirmation gate or rollback mechanism shown; sensitive hiding is opt-in and arbitrary free-form secrets may evade it; dependencies have lower bounds but no security audit or lock strategy is shown; effects include direct router changes, backup restoration, and file upload; publisher identity is unverified.
Installation docs specify Python versions, transports, timeouts, health checks, and a locked source-install path; CI configuration describes lockfile drift checks and test workflows. Deductions: the supplied README, installation guide, and examples broadly align but do not establish that every described behavior matches implementation; concrete runtime failure messages and recovery guidance are sparse.
The project targets users connecting AI assistants to MikroTik RouterOS over MCP, covers single devices and inventories, stdio and HTTP transports, and documents read-only and deployment settings. Deductions: scenarios are broad, but explicit guidance on when tools should trigger or not be called is thin; capability boundaries are conveyed mainly through security and configuration notes rather than a fuller operational boundary description.
Documentation is well organized and provides detailed installation paths, prerequisites, CLI and environment settings, Docker, and proxy guidance. The MIT license text is complete, and security considerations and known limitations are explicit. Deductions: examples are extensive but include passwords and high-impact restore, log-clearing, and configuration actions without per-example risk prompts; versioning is release-driven, while no changelog is supplied; contribution and security-reporting paths exist, but maintenance responsibility and response commitments are only partly concrete.
The server connects natural-language assistants to RouterOS management, including network configuration, monitoring, backups, and device inventories, giving it clear automation value. Deductions: examples focus on tool invocations and do not show result formats, validation steps, or readable post-change summaries; direct network-device management has a high cost of configuration error, and no default confirmation or transactional recovery is shown.
Many security and installation claims can be cross-checked among the supplied security policy, installation guide, project metadata, and CI configuration; the docs also distinguish configuration facts from deployment advice. Deductions: implementation source and lockfile contents were not supplied, so badges and external audit claims cannot be verified from these files, and some described capabilities cannot be traced item by item to implementation.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- HTTP transports have no built-in caller authentication and bind to 0.0.0.0 by default; use them only on trusted networks with access restrictions or external authentication.
- SSH host keys are automatically accepted when unknown; assess the man-in-the-middle risk before production use.
- Read-only mode limits the MCP tool catalogue but does not enforce RouterOS permissions; use a restricted device account.
- Sensitive-data hiding is off by default and cannot reliably identify arbitrary secrets in free-form comments or logs.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
v0.26.2 Release v0.26.2
Release notes excerpt
Changes in v0.26.2 This release was automatically created by GitHub Actions. PyPI package version: 0.26.2.0 Git tag: v0.26.2 Commit: e0a90d33324356567b536ba51fbac17c1791980dView on GitHub ↗ -
v0.26.1 Release v0.26.1
Release notes excerpt
Changes in v0.26.1 This release was automatically created by GitHub Actions. PyPI package version: 0.26.1.0 Git tag: v0.26.1 Commit: a5216cdaab2ac0b7f4b9065d2cc699a205c92b52View on GitHub ↗ -
v0.26.0 Release v0.26.0
Release notes excerpt
Changes in v0.26.0 This release was automatically created by GitHub Actions. PyPI package version: 0.26.0.0 Git tag: v0.26.0 Commit: 2377ecb015ebe5d389ac7748b6e00bb98a566b3eView on GitHub ↗