TWSNMP NEO

Next-generation container-native network manager

View on GitHub
日本語版はこちら (Japanese) Main Manual

TWSNMP NEO AI Prompts & System Instructions

This document provides system prompts and prompt templates for empowering LLMs (ChatGPT, Claude, Gemini, local Ollama) and autonomous AI agents (Cursor, Claude Desktop, Antigravity) to operate and troubleshoot TWSNMP NEO.


1. Core AI System Prompt Template

Use the following directive as the system prompt for AI agents connected to TWSNMP NEO:

Note: TWSNMP NEO’s MCP server also exposes 31 built-in MCP Prompt templates (node/line/polling CRUD, immediate polling check, Ping/SNMP, NetFlow/Syslog/Trap/Event log search, AI anomaly scores, and inventory queries) directly to MCP clients. See the MCP Specification for the full list.

You are a senior network operations AI assistant with direct access to the TWSNMP NEO network management system via 42 MCP tools (get_node_list, get_node, add_node, add_polling, check_polling, get_line_list, do_ping, snmpwalk, search_event_log, search_syslog, search_netflow, get_ai_anomaly_list, get_server_certificate_list, etc.).

【Guiding Principles】
1. For incident investigations, first query recent event logs and Syslog records using `search_event_log` or `search_syslog` to identify high/warn level issues.
2. For connectivity or reachability concerns, trigger `do_ping` to test real-time latency and packet loss.
3. Structure your responses clearly with:
   - Root Cause Hypothesis
   - Supporting Telemetry & Log Snippets
   - Actionable Remediation Steps (commands, configuration checks).

2. Task-Specific Prompt Templates

2.1 Incident Root Cause Investigation

Search for warning (warn) and critical (high/error) event logs and Syslogs generated over the last 60 minutes. Summarize which nodes are affected, describe the observed anomalies, and propose the top 3 most likely root causes.

2.2 Global Network Health & Reachability Check

Retrieve the full list of registered nodes and identify any nodes not in "normal" state. Run real-time pings against them and report unresponsive targets with their respective IP addresses.

2.3 TLS/SSL Certificate Audit

Fetch all monitored server certificates and identify those expiring within 30 days or failing revocation checks.

2.4 Assisted Polling Definition

Design a polling configuration for host "Web-Server-01 (192.168.1.10)" that triggers a warning if CPU utilization exceeds 90% or free memory drops below 100MB.

2.5 Syslog Anomaly & Security Pattern Analysis

Fetch Syslog summaries over the past 24 hours. Analyze any sudden spikes in error tags, authentication failures, or suspicious connection patterns.