Loading ToolBench
The first MCP server for network equipment — let AI agents talk to Nokia SR OS, Cisco IOS-XR, and Juniper Junos routers
All 11 tools have descriptions (50-150 chars, within baseline range) and single required parameter 'device' with description. However, schemas lack critical structure: no explicit JSON Schema with type/description pairs visible in registration; parameters appear as bare strings without formal constraints. Tool names follow verb_noun pattern (show_*) but descriptions are generic, they state WHAT without WHEN or WHY. No output schemas documented. Error handling returns raw exception strings ('Error: {e}') without recovery guidance. No tool annotations (readOnlyHint, etc.). Composition is sound (each tool does one thing), but parameter validation and error classification are absent.
Show the ARP table on a network device. Returns IP-to-MAC address mappings for all entries in the ARP cache.
Show detailed BGP neighbor information on a network device. Returns per-neighbor details including state, AS number, messages sent/received, and hold time.
Show BGP neighbor summary on a network device. Returns the BGP peering table with neighbor states, prefixes received, and session uptime. Works across all supported vendors.
Show detailed statistics for a specific interface. Returns counters, errors, CRC, drops, utilization, MTU, and speed for the specified interface.
Show interface status summary on a network device. Returns all interfaces with admin/operational status, speed, and description.
Show Link Aggregation Group (LAG/Port-Channel/Bundle) status. Returns LAG member ports, their states, and aggregate bandwidth.
No formal JSON Schema with type/description pairs visible in tool registration. Parameter 'device' appears as bare string without enum, pattern, or constraint documentation. LLMs cannot validate device names or suggest valid options.
No output schemas documented. Tools return raw string output from network devices without structure. LLMs cannot parse results, extract fields for downstream calls, or validate completeness.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | <=2025-11-25 | v2 |
Show LLDP neighbor discovery information. Returns connected neighbors with their system name, port ID, and capabilities — useful for topology discovery.
Show the MAC address table on a network device. Returns MAC addresses, VLANs, and associated interfaces.
Show MPLS Label Switched Paths on a network device. Returns LSP status, tunnel endpoints, and label bindings.
Show OSPF neighbor adjacencies on a network device. Returns OSPF neighbor states, interface associations, and dead intervals.
Show the IP routing table on a network device. Returns all routes with protocol, next-hop, metric, and preference.
Error handling returns raw exception strings ('Error: {e}') without recovery guidance. LLMs cannot determine if error is retryable, user-fixable, or fatal. No actionable next steps provided.
Descriptions lack WHEN/WHY context. 'Show BGP neighbor summary' states WHAT but not when to call it vs show_bgp_neighbors, or what prerequisites exist. LLMs cannot distinguish between similar tools.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are read-only but this is not declared. Agents cannot infer safety properties or plan retry strategies.