MCP server for system monitoring — CPU, memory, disk, network, and process management
Server has 5 tools with clear verb-noun naming (get_*, find_*) and detailed descriptions (150-300 chars each). All tools are read-only with explicit risk annotations. However, input schemas are incomplete: get_system_overview, get_disk_usage, get_network_info, and find_process lack visible parameter type definitions in the source. get_top_processes has proper schema with type, description, default, and range constraints. Output schemas are not documented, descriptions mention Markdown tables/reports but no structured field definitions. Error handling is minimal; no recovery guidance or actionable error messages visible. Composition is sound (each tool does one thing), but missing pagination for potentially large result sets (e.g., get_top_processes capped at 50 but no offset/limit for discovery).
Search for running processes by name using case-insensitive partial matching. Use this to locate specific processes — for example, find_process("chrome") returns all Chrome-related processes. Use get_top_processes instead when you want to see the highest resource consumers regardless of name. After finding a process, you can use its PID with kill_process to terminate it. This is a read-only operation with no side effects.
Show detailed disk usage for all mounted partitions with I/O statistics. Use this for disk space analysis — identifying full partitions, comparing filesystem usage, or checking I/O throughput. For a quick disk summary as part of overall system health, use get_system_overview instead. This is a read-only operation with no side effects. Returns a Markdown table with columns: Mount, Device, Total, Used, Free, Usage%, FS type. Also includes cumulative disk I/O since boot (total bytes read/written). Partitions requiring elevated privileges are silently skipped.
Show all network interfaces with IP addresses, link speed, status, and traffic counters. Use this to check network connectivity, find the machine's IP addresses, or investigate network throughput. Not suitable for packet-level analysis or firewall rule inspection. This is a read-only operation with no side effects. Returns a Markdown report grouped by interface. Each interface shows: UP/DOWN status, link speed in Mbps, IPv4/IPv6 addresses, and cumulative bytes sent/received since boot. Loopback and virtual interfaces are included.
Retrieve a full system health snapshot: CPU usage, core count, memory, swap, disk usage, and uptime. Use this as the first tool when diagnosing system performance issues or answering questions like "why is my machine slow?" or "how much disk space is left?". For deeper investigation, follow up with get_top_processes, get_disk_usage, or get_network_info. This is a read-only operation with no side effects. Takes ~0.5 seconds due to CPU sampling. Returns a Markdown report with sections: CPU, Memory, Swap, and Disks. Each section shows current usage as absolute values and percentages. Disk partitions that require elevated privileges are silently skipped.
Four of five tools lack visible input schema definitions. get_system_overview, get_disk_usage, get_network_info have no parameters declared in source, and find_process has only a name parameter with type string but no constraints (min/max length, pattern).
Output schemas are not documented. Descriptions state 'Returns a Markdown report' or 'Returns a Markdown table' but do not specify field names, types, or structure. LLMs cannot plan downstream tool calls or extract specific data without knowing the response schema.
No error handling or recovery guidance visible. Descriptions mention 'Partitions requiring elevated privileges are silently skipped' and 'Processes that exit during enumeration are skipped' but no tool returns error details or suggests next steps if a call fails.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
List the top resource-consuming processes sorted by CPU or memory usage. Use this to identify which processes are consuming the most resources. Use sort_by="memory" (default) to find memory hogs, or sort_by="cpu" to find CPU-intensive processes. Use get_system_overview first for the big picture, then this tool to drill down into specific processes. To search for a specific process by name, use find_process instead. This is a read-only operation with no side effects. When sorting by CPU, takes ~0.5 seconds for accurate sampling; memory sorting is instant. Returns a Markdown table with columns: PID, Name, CPU%, Mem%, RSS, Status. Processes that exit during enumeration or require elevated access are skipped.
find_process parameter 'name' lacks constraints. No min/max length, no regex pattern, no guidance on case sensitivity (description says 'case-insensitive' but schema does not enforce it).
get_top_processes accepts limit 1-50 with auto-clamping, but no pagination (offset/cursor) for discovering processes beyond top 50. If an agent needs to find a specific process not in top 10, it must call find_process separately.