[
  {
    "slug": "mcp-server-management",
    "date": "2026-03-31",
    "author": "ManageLM Team",
    "tags": ["mcp", "architecture", "security"],
    "image": "blog2.webp",
    "title": "Why MCP Changes Everything for Server Management",
    "summary": "The Model Context Protocol (MCP) by Anthropic gives AI tools a structured, authenticated way to interact with external systems. For server management, this is transformative — it replaces fragile scripts and manual SSH sessions with secure, auditable, natural-language operations. Here's how ManageLM leverages MCP to make infrastructure management safer and more accessible.",
    "content": "<p>If you manage servers, you know the drill: SSH into a box, run a sequence of commands you half-remember, pipe the output through grep, hope nothing breaks. Or you maintain a growing collection of Ansible playbooks and Bash scripts that nobody else on the team fully understands.</p><p>There's a better way — and it starts with <strong>MCP</strong>.</p><h3>What Is MCP?</h3><p>The <strong>Model Context Protocol</strong> (MCP) is an open standard created by Anthropic that defines how AI assistants like Claude interact with external tools and data sources. Instead of the AI generating raw shell commands and hoping for the best, MCP provides a structured interface: the AI sends well-defined requests to an MCP server, which validates them, routes them, and returns structured responses.</p><p>Think of MCP as the difference between handing someone a keyboard to your server and giving them a controlled API. The AI gets the <em>capabilities</em> it needs without getting <em>unrestricted access</em>.</p><h3>MCP + Server Management: A Natural Fit</h3><p>Server management is one of the most compelling use cases for MCP, because it combines two things that rarely go together: <strong>the need for flexibility</strong> (every server is different, every situation is unique) and <strong>the need for strict security</strong> (one wrong command can take down production).</p><p>MCP solves this tension elegantly:</p><ul><li><strong>Structured tool definitions</strong> — Each ManageLM skill (packages, services, firewall, databases, etc.) is exposed as a set of MCP tools with defined parameters. The AI can't invent operations that don't exist.</li><li><strong>Authentication and authorization</strong> — Every MCP request carries the user's identity through OAuth 2.0. The portal checks RBAC permissions before dispatching anything. Different team members can have different access levels.</li><li><strong>Auditability</strong> — Every MCP call is logged with full context: who requested it, what was executed, what changed. This is built into the protocol flow, not bolted on after the fact.</li><li><strong>Multi-server targeting</strong> — MCP's structured request format makes it natural to target operations at specific servers, groups, or entire fleets. \"Update packages on all staging servers\" becomes a single MCP call that fans out to multiple agents.</li></ul><h3>How ManageLM Uses MCP</h3><p>ManageLM is built as an MCP server that Claude connects to directly. When you tell Claude <em>\"check disk usage on web-prod-01\"</em>, here's what happens:</p><ol><li><strong>Claude parses your intent</strong> and maps it to the appropriate ManageLM MCP tool (<code>system</code> skill, <code>disk_usage</code> operation).</li><li><strong>The MCP request hits the ManageLM portal</strong>, which authenticates your session (OAuth 2.0), verifies you have permission to access web-prod-01, and validates the requested skill.</li><li><strong>The portal dispatches the task</strong> over a secure WebSocket to the agent running on web-prod-01.</li><li><strong>The agent uses a local LLM</strong> (Ollama, running in your infrastructure) to interpret the task and generate the exact shell commands needed.</li><li><strong>Every generated command is validated</strong> against the skill's allowlist before execution. If the LLM hallucinates a dangerous command, it's blocked in code.</li><li><strong>Results flow back</strong> through the same chain: agent → portal → Claude → you, in natural language.</li></ol><p>The entire flow is secured at every layer. The AI is powerful but <em>untrusted by design</em> — it proposes commands, but the allowlist enforces what actually runs.</p><h3>Why Not Just Give the AI SSH Access?</h3><p>You could give an AI tool an SSH key and let it run whatever it wants. Some products do this. It's a terrible idea, and here's why:</p><ul><li><strong>LLMs hallucinate.</strong> A model might generate <code>rm -rf /</code> when it meant <code>rm -rf /tmp/cache</code>. Without validation, that command runs.</li><li><strong>Prompt injection is real.</strong> If the AI processes untrusted input (log files, user data), an attacker could manipulate the AI into running malicious commands.</li><li><strong>No audit granularity.</strong> SSH logs show that a key connected and ran commands, but don't capture the intent, the user who initiated it, or the AI reasoning behind it.</li><li><strong>No permission scoping.</strong> SSH gives all-or-nothing access. MCP through ManageLM gives skill-scoped, role-based access with per-command validation.</li></ul><p>MCP's structured protocol makes it possible to build AI-powered server management that's actually <em>more secure</em> than traditional approaches, not less.</p><h3>Beyond Claude: MCP as an Open Standard</h3><p>Because MCP is an open protocol, ManageLM isn't locked to a single AI provider. Today it works with Claude (MCP's native home), and the same architecture extends to other interfaces — ChatGPT via a GPT plugin, VS Code via a Copilot extension, Slack for alerts and approvals, and n8n for automation pipelines. The MCP server is the single source of truth for authentication, authorization, and audit, regardless of which AI or interface triggers the operation.</p><h3>Getting Started</h3><p>ManageLM is <strong>free for up to 10 agents</strong> with every feature included. Install the <a href=\"https://www.managelm.com/plugins/claude.html\" target=\"_blank\">Claude MCP extension</a>, connect your servers, and start managing your infrastructure in natural language — with security enforced at every layer.</p><p>The future of server management isn't more YAML files or longer Bash scripts. It's a conversation with an AI that actually understands your intent, scoped by security policies that actually enforce your rules. That's what MCP makes possible, and that's what ManageLM delivers.</p>"
  },
  {
    "slug": "introducing-managelm",
    "date": "2026-03-25",
    "author": "ManageLM Team",
    "tags": ["announcement", "product"],
    "image": "blog1.webp",
    "title": "Introducing ManageLM: AI-Powered Server Management",
    "summary": "Meet ManageLM — the platform that lets you manage your entire Linux and Windows infrastructure using natural language. Built on Claude's MCP protocol with local LLM execution, command allowlisting, and zero inbound ports.",
    "content": "<p>We're excited to introduce <strong>ManageLM</strong>, a new way to manage your servers. Instead of writing scripts, memorizing commands, or navigating complex dashboards, you simply tell an AI what you need — and it gets done.</p><h3>What is ManageLM?</h3><p>ManageLM is an AI-powered server management platform. It connects Claude (via the Model Context Protocol) to lightweight agents running on your Linux and Windows servers. You describe tasks in plain English — <em>\"restart nginx on all staging servers\"</em>, <em>\"check disk usage\"</em>, <em>\"update packages\"</em> — and ManageLM handles the rest.</p><p>But unlike generic AI tools, ManageLM was built with security as the architecture, not an afterthought. Every command generated by the AI is validated against an explicit allowlist before execution. The AI is untrusted by design.</p><h3>How It Works</h3><p>The platform has three layers:</p><ul><li><strong>Claude + MCP</strong> — You talk to Claude in natural language. Claude sends structured requests to the ManageLM portal via MCP.</li><li><strong>Portal</strong> — Authenticates your identity, checks RBAC permissions, validates the target agent, and dispatches the task over a secure WebSocket channel.</li><li><strong>Agent</strong> — A lightweight process on your server that uses a local LLM (Ollama) to interpret the task, generate commands, validate each one against the skill's allowlist, and execute. Your data never leaves your network.</li></ul><h3>Security First</h3><p>ManageLM was designed for teams that take security seriously:</p><ul><li><strong>Command allowlisting</strong> — Every skill defines exactly which commands are permitted. The AI cannot run anything outside this list.</li><li><strong>Local LLM execution</strong> — Task interpretation runs on your infrastructure via Ollama. Passwords, configs, and logs never leave your network.</li><li><strong>Zero inbound ports</strong> — Agents connect outward via WebSocket. Your servers never expose a port.</li><li><strong>Ed25519 signed messages</strong> — Every portal-to-agent message is cryptographically signed. Tampered messages are rejected.</li><li><strong>Kernel sandbox</strong> — Optional Landlock + seccomp confinement for defense-in-depth.</li></ul><h3>Get Started</h3><p>ManageLM is <strong>100% free</strong> for up to 10 agents — every feature included, no credit card required. You can use our managed cloud at <a href=\"https://app.managelm.com/register\" target=\"_blank\">app.managelm.com</a>, or self-host with Docker for full data sovereignty.</p><p>We can't wait to see what you build with it.</p>"
  }
]