Model Context Protocol server for IBM MQ queue manager administration and message operations
IBM MQ MCP has 15 tools with complete input schemas and consistent descriptions. All tools follow verb_noun naming (get_, list_, create_, delete_, update_, clear_, browse_, put_, start_, stop_). Descriptions are present and adequate (50-100 chars), explaining WHAT each tool does. However, parameter descriptions are minimal, most lack detail on constraints, formats, or when to use them. Output schemas are not documented in the visible code. Error handling guidance is absent. No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels. The connection object is repeated across all 15 tools, creating maintenance burden and parameter bloat. Missing: pagination guidance for list_queues and list_channels, confirmation patterns for destructive ops (delete_queue, clear_queue, delete_channel), and actionable error messages.
Browse queue messages without consuming them.
Clear messages from a queue using PCF.
Create a channel in one of the supported types: SVRCONN, SDR, RCVR, CLNTCONN.
Create a local queue with a curated set of attributes.
Delete a channel by name.
Delete a queue by name.
Get a single channel with merged definition and status data.
Connection object repeated in all 15 tools. Bloats every parameter list and creates maintenance burden. Should be extracted to server-level configuration or a single 'connection' parameter with reusable schema.
Destructive tools (delete_queue, clear_queue, delete_channel) lack confirmation patterns or dry-run support. No error guidance on what happens when deletion fails. Agents may accidentally destroy data.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
Get a single queue with merged definition and status data.
Get queue manager properties and status.
List channels and merge definition and status data.
List queues and merge definition and status data.
Put a test message to a queue using text or base64 payloads.
Start a channel by name.
Stop a channel by name.
Update a curated set of queue attributes.
List tools (list_queues, list_channels) lack pagination parameters (limit, offset, cursor) and result count documentation. Large queue/channel lists will blow context window.
Output schemas not documented in visible code. LLMs cannot plan downstream calls or extract required fields (e.g., queue_id for chaining). Responses appear to be JSON but structure is opaque.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. Agents cannot distinguish safe reads from destructive writes without parsing descriptions.