One workbench for live MQTT clients, traffic replay, broker benchmarks, fault injection, PCAP analysis, and bounded agent workflows.
Try Grid, Stream, Topics, and Sparkplug, then open the message composer or AI Guide.
Serious MQTT work needs more than a publish and subscribe form.
Paste logs and screenshots into chat, explain the broker state, then repeat every diagnosis and action by hand.
Rebuild multi-client flows one message at a time, losing the original timing, sessions, and ACK behavior.
Set up generators, tune the host, wire up metrics, inject faults, and switch tools before getting a useful result.
Let MCP-compatible agents inspect traces, run diagnostics, compare benchmarks, and safely operate the MQTT workbench.
Turn live traces, CSV, or PCAP into timed multi-client scenarios with dry runs, ACK checks, and delivery verification.
Launch built-in benchmarks, inject MQTT-aware network faults, and turn PCAP captures into ranked root causes.
Go beyond MQTT pub/sub with Insights that surfaces behavior risks, traffic patterns, and the evidence behind them.
Grid shows what happened, Stream shows when, Topics shows where, and Sparkplug shows what it means for each device.
Send and trace your first MQTT message
Recreate how multiple clients connect, subscribe, and publish, then run the same flow on your broker.
Keep the original timing, redirect topics to a test environment, and verify responses and message delivery after every run.
Reproduce a hard-to-catch MQTT bug with Replay
Simulate concurrent MQTT clients, control connection and publish rates, and see where performance begins to change.
Catch problems with live metrics, then review the full run in throughput, latency, and connection charts.
Run your first MQTT benchmark in 5 minutes
Place Mqttable between your MQTT clients and broker, then make the connection slow, congested, unreliable, or completely unavailable. See which messages arrive, how clients reconnect, and whether the system recovers before the same failure happens in production.
Build repeatable failure scenarios, inspect the results in Trace, and verify cleanup and recovery before production.
Inject and verify MQTT latency safely
Sixty checks examine connection lifecycle, message delivery, subscription behavior, protocol correctness, MQTT 5 properties, and traffic health.
Each flagged result points to the client, session, and packet evidence you need to investigate the failure.
Find MQTT root-cause evidence in a PCAP
See how Mqttable has been running over the last hour.
Review CPU and memory usage, file descriptor and thread counts, uptime, scheduler activity, and recent events in one place.
Read Runtime metrics and trends
A collection of small tools built into Mqttable.
Decode JWTs, generate QR codes, compress images, compare text, inspect gzip data, hash files, and convert timestamps.
Explore the built-in engineering tools
Stop asking what failed. With Mqttable MCP, your agent can inspect the runtime, operate MQTT, and verify the result.
Create connections, publish and subscribe, run bounded benchmarks, inject network faults, inspect traces, and analyze PCAPs through Mqttable MCP.
MCP Monitor shows each tool’s calls, success rate, P95 latency, and last use, alongside recent issues, Plans, Apply, and Verification. Writes stay behind local authorization, and raw secrets stay out of MCP.
Nine stable tools reveal exact actions, schemas, and resources only when needed. The full capability catalog never floods the prompt, leaving more context for MQTT traces and diagnosis.
Compare where each app runs, which MQTT connections it supports, which testing and troubleshooting tasks it can handle, and how it performed on the same Apple M3 Mac.
| What you can compare | Mqttable1.2.0 | MQTTX1.13.0 | MQTT Explorer0.4.0-beta.6 | MQTT.fx5.14.2 |
|---|---|---|---|---|
| Installation and connections | ||||
| How the app is builtThe main desktop technology used by each app. | Built with Tauri 2.11.5 and Phoenix 1.8.13. | Built with Electron 39.2.7. | Built with Electron 29.2.0. | Built with JavaFX 21.0.1 on JDK 21.0.1. |
| Where you can install itSupported desktop systems and computer types. | Available for macOS 15+ on Apple Silicon and Linux. |
| Available for Windows, macOS, and Linux. | Available for Windows, macOS, and Linux. |
| MQTT versions and connection types |
|
|
|
|
| Testing and troubleshooting | ||||
| Follow messages without switching toolsSee live messages, search the history, and read their contents. | Included in the app See live topics, search message history, and read text, JSON, or raw data without switching apps. | Included in the app Browse live topics, inspect message contents, compare changes, and open JSON in a tree view. | Included in the app Browse live topics, search and filter messages, view each topic’s history, compare changes, and chart numbers. | Included in the app Browse topics, filter received messages, view history, and add decoders for different message formats. |
| Reproduce traffic without writing scriptsTurn saved MQTT traffic into a test you can run again. | Included in the app Turn saved MQTT traffic, a CSV file, or a network capture into a test you can run again. | Separate tool or partial support Scripts and a separate command-line app can repeat send-and-receive steps. The sources checked did not show replay of saved traffic from several clients. | Not found in the sources checked | Separate tool or partial support Built-in scripts can repeat selected send-and-receive steps. The sources checked did not show replay of traffic previously captured from several clients. |
| Create test traffic in the same appRun simulated clients and see how the MQTT broker responds. | Included in the app Choose how many simulated clients to run and how often they send messages, then watch speed, delay, failures, connection health, and computer usage. | Separate tool or partial support A separate command-line app can test connections, sending, and receiving; this is not included in the desktop app. | Not found in the sources checked | Separate tool or partial support Scripts can create message traffic. The sources checked did not show a standard test that also tracks message delay and computer usage. |
| Test bad networks without changing your brokerTry delay, limited bandwidth, packet loss, timeouts, and disconnects. | Included in the app Send test traffic through the app and add delay, bandwidth limits, packet loss, timeouts, or disconnects. | Not found in the sources checked | Not found in the sources checked | Not found in the sources checked |
| Start with findings, not raw packetsOpen a saved network traffic file (PCAP) and find likely MQTT problems. | Included in the app The app highlights likely MQTT problems in a network capture and links each finding to the supporting packets. | Not found in the sources checked | Not found in the sources checked | Not found in the sources checked |
| Use automation without giving up controlRepeat MQTT tasks with scripts, command-line tools, or AI assistants. | Included in the app Connect a compatible AI assistant while sensitive connection details stay hidden and changes remain permission-controlled. | Separate tool or partial support Desktop scripts and a separate command-line app can automate messages. The sources checked did not show a built-in AI interface for the desktop app. | Not found in the sources checked | Separate tool or partial support Built-in scripts can automate MQTT messages. The sources checked did not show a command-line or AI assistant interface. |
MacBook Air with Apple M3 and 16 GB RAM. All four apps were measured together on macOS 26.6.2 on Sep 2, 2026. Each result uses one warm-up and five measured launches with product order rotated between rounds; Footprint and RSS use their own five-run medians.
| Metric | Mqttable1.2.0 | MQTTX1.13.0 | MQTT Explorer0.4.0-beta.6 | MQTT.fx5.14.2 |
|---|---|---|---|---|
| Warm launch to usable workspace | 1.35 s | 2.46 s | 2.05 s | 3.80 s |
| Memory footprint 30 seconds after ready | 339.3 MiB | 321.4 MiB | 191.3 MiB | 479.7 MiB |
| Resident set 30 seconds after ready |
|
|
|
|
| Installed app size | 68 MiB | 317 MiB | 269 MiB | 434 MiB |
The main desktop technology used by each app.
Mqttable
Built with Tauri 2.11.5 and Phoenix 1.8.13.
MQTTX
Built with Electron 39.2.7.
Supported desktop systems and computer types.
Mqttable
Available for macOS 15+ on Apple Silicon and Linux.
MQTTX
Mqttable
MQTTX
See live messages, search the history, and read their contents.
Mqttable
See live topics, search message history, and read text, JSON, or raw data without switching apps.
MQTTX
Browse live topics, inspect message contents, compare changes, and open JSON in a tree view.
Turn saved MQTT traffic into a test you can run again.
Mqttable
Turn saved MQTT traffic, a CSV file, or a network capture into a test you can run again.
MQTTX
Scripts and a separate command-line app can repeat send-and-receive steps. The sources checked did not show replay of saved traffic from several clients.
Run simulated clients and see how the MQTT broker responds.
Mqttable
Choose how many simulated clients to run and how often they send messages, then watch speed, delay, failures, connection health, and computer usage.
MQTTX
A separate command-line app can test connections, sending, and receiving; this is not included in the desktop app.
Try delay, limited bandwidth, packet loss, timeouts, and disconnects.
Mqttable
Send test traffic through the app and add delay, bandwidth limits, packet loss, timeouts, or disconnects.
MQTTX
Open a saved network traffic file (PCAP) and find likely MQTT problems.
Mqttable
The app highlights likely MQTT problems in a network capture and links each finding to the supporting packets.
MQTTX
Repeat MQTT tasks with scripts, command-line tools, or AI assistants.
Mqttable
Connect a compatible AI assistant while sensitive connection details stay hidden and changes remain permission-controlled.
MQTTX
Desktop scripts and a separate command-line app can automate messages. The sources checked did not show a built-in AI interface for the desktop app.
MacBook Air with Apple M3 and 16 GB RAM. All four apps were measured together on macOS 26.6.2 on Sep 2, 2026. Each result uses one warm-up and five measured launches with product order rotated between rounds; Footprint and RSS use their own five-run medians.
Mqttable
MQTTX
Capability rows use current first-party sources and the exact versions shown. “Not documented” is an evidence boundary, not a claim that a capability cannot exist.
Mqttable: 1.2.0. Competitors used official Apple Silicon installers.
Ready means: Mqttable returned 200 from /__desktop/ready and exposed Connections plus Connect Now; MQTTX showed Connections and an enabled New Connection action; MQTT Explorer showed its Connections form; MQTT.fx passed Continue Without License and showed Namespace Explorer, Connect, and Publish. Thirty seconds later, the runner captured RSS and then a multi-PID macOS footprint. No broker connection or workload was used.
Every product used isolated app-data and reused the clean state created by its warm-up. All four apps were measured in the same run, with product order rotated between rounds. MQTT Explorer was ad-hoc re-signed after its bundled signature failed validation. Footprint de-duplicates multiply mapped objects within the attributed PID set; RSS may count shared pages more than once. Mqttable WebContent and Networking were tied to its bundle data paths, while its paired GPU process was identified by the launch/quit lifecycle.
Free local workbench
Connect, inspect, replay, benchmark, proxy, analyze PCAPs, monitor runtimes, and use local MCP. Free covers everyday MQTT engineering; upgrade only when an investigation needs more scale, longer runs, or full evidence.
The complete local workbench for everyday MQTT engineering, with clear investigation limits.
$0/ forever
Start with common questions about setup, connection management, and MQTT troubleshooting. Still need help? support@mqttable.com
