How does this skill differ from simply asking a standard AI model for IBM MQ help?

This skill automates the analysis of complex MQ issues, such as 2035/2059 reason codes and cluster repository sync errors, by pulling live data through the MQ-Sentinel MCP server rather than relying on static training data.

What specific infrastructure or software is required to use this skill?

The skill requires the MQ-Sentinel MCP server (available via PyPI) to be installed and configured on your host; it uses this bridge to safely execute diagnostic commands and fetch real-time MQSC evidence.

Can the diagnostic tool perform administrative changes or fix the issues it identifies?

No, this skill is designed with a "read-only" philosophy; it identifies root causes and suggests remediation MQSC commands, but it does not execute changes or configuration updates to your queue managers.

Does the diagnostic skill support mainframe environments like IBM MQ for z/OS?

Yes, the skill is fully compatible with IBM MQ for z/OS, including specialized health checks for Queue Sharing Groups (QSG), Coupling Facility structures, and buffer pool utilization.

How does the skill handle sensitive data when inspecting the Dead Letter Queue?

The skill follows a "privacy-first" approach for Dead Letter Queue analysis, inspecting headers and reason codes to suggest reprocessing steps without accessing or displaying the actual message payload.

IBM MQ Diagnostics (MQ-Sentinel)

Example session with this skill installed

Channel APP.SVRCONN on DEMO_QM keeps going into RETRYING. Can you check our channels and tell me what's wrong?

Read your context and instructions
Compiled the ibm mq diagnostics

I ran diagnose_failed_channels on DEMO_QM with MQ-Sentinel (read-only, demo data).

DEMO_QM: 3 channel problems. MQ 9.4.0.0, Native HA. 4 channels examined.

Severity	Channel	Finding
🔴 CRITICAL	INDOUBT.RCVR	Channel is in doubt
🟠 HIGH	APP.SVRCONN	MQRC 2035 (NOT_AUTHORIZED)
🟠 HIGH	TO.PARTNER	Connection error (reason 2009)

🟠 APP.SVRCONN: MQRC 2035 (NOT_AUTHORIZED)

Root cause: the connecting principal failed authorization. Common causes: a CHLAUTH BLOCKUSER rule blocked the user, the resolved MCAUSER lacks +connect/+inq on the queue manager, or CONNAUTH credentials failed.

Evidence: status RETRYING, remote QM REMOTE_QM. Confidence: High.

Confirm it (read-only):

DISPLAY CHLAUTH('APP.SVRCONN') MATCH(RUNCHECK) ALL
DISPLAY CHSTATUS('APP.SVRCONN') ALL
DISPLAY QMGR CONNAUTH
DISPLAY AUTHINFO(*) ALL

IBM docs: 2035 (07F3) MQRC_NOT_AUTHORIZED

Possible fixes, depending on what the checks show (for you to review and run in a change window):

MCAUSER missing permissions: SET AUTHREC PRINCIPAL('mcauser') OBJTYPE(QMGR) AUTHADD(CONNECT, INQ). Replace mcauser with the value from DISPLAY CHSTATUS MCAUSER.
A BLOCKUSER rule matching the wrong user: narrow its USERLIST to only the principals you mean to block.
CONNAUTH rejecting the credentials: check the AUTHINFO configuration, then REFRESH SECURITY TYPE(CONNAUTH).

🔴 INDOUBT.RCVR: channel in doubt

Root cause: uncommitted batches can't resolve on their own; manual resolution is needed after checking the remote QM.

Confirm: DISPLAY CHSTATUS('INDOUBT.RCVR') ALL CURLUWID LSTLUWID

⚠️

Only after confirming the batch state on the remote QM: RESOLVE CHANNEL('INDOUBT.RCVR') ACTION(COMMIT), or ACTION(BACKOUT). Choosing wrongly duplicates or loses messages.

🟠 TO.PARTNER: connection error (2009)

Root cause: network or TLS handshake failure to the remote endpoint.

Confirm:

DISPLAY CHSTATUS('TO.PARTNER') ALL
DISPLAY CHANNEL('TO.PARTNER') CONNAME XMITQ SSLCIPH
DISPLAY LSSTATUS(*) ALL
PING CHANNEL(TO.PARTNER)

IBM docs: AMQ9202E: Remote host not available

Next steps

Run the read-only checks above, starting with APP.SVRCONN's CHLAUTH check.
Decide on INDOUBT.RCVR with your messaging team before resolving anything.
Nothing was changed on the queue manager. MQ-Sentinel only reads.

Connects securely to your tools. The creator never sees your data.

About this skill

Ask your agent why a channel is RETRYING, what MQRC 2035 means on your queue manager, why the dead-letter queue is growing, or whether your cluster or Native HA setup is healthy. You get a cited diagnosis: the evidence, the root cause, the DISPLAY commands to confirm it, links to the matching IBM Documentation pages, and possible fixes for you to review.

The skill drives MQ-Sentinel, a free, read-only MCP server for IBM MQ:

9 tools: channels, dead-letter queue (headers only, bodies never read), cluster, Native HA, RDQM, multi-instance, z/OS queue-sharing groups, and a full health check with an executive summary
Read-only by design: only allowlisted DISPLAY commands reach your queue managers; suggested changes come back as text for you to run
Every IBM documentation link it cites is checked daily
Try it in demo mode with no IBM MQ, then connect your own queue managers, or a central team server with SSO

Includes ready-to-copy configs for Claude Code, Claude Desktop, Cursor and VS Code, a least-privilege MQ setup guide (setmqaut, CHLAUTH), and a reason-code reference.

Requires the free MQ-Sentinel MCP server: https://pypi.org/project/mq-sentinel/ · Source: https://github.com/pramodreddyboddu/mq-sentinel

How to use

Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

Reviews

No reviews yet

Be one of the first to try it. Every listed skill passes our trust checks below.

Security scanned

Passed our 8-point scan before listing

Fresh listing

Recently published to Agensi

Free forever

No account required to browse

IBM MQ Diagnostics (MQ-Sentinel)