Mqttable

AI-NATIVE MQTT DEBUGGING

The MQTT workbench for humans and AI agents.

Capture connections, traces, packet data, replays, benchmarks, and runtime metrics as structured evidence—so engineers and agents can find the cause and verify the fix.

See AI troubleshootDownload desktop

Local-first desktop workbench. Agent collaboration runs through CLI and MCP—there is no built-in autonomous model.

One evidence chain, readable by humans and agents

ConnectionsTracePCAPReplayProxyBenchMonitor→ CLI / MCP → Agent

A real troubleshooting path

Why does this QoS 1 publish keep retrying?

See how an agent can move from bounded evidence to a masked plan, stop for human confirmation, and verify the result in the same workspace.

Product capability demo

This fixed walkthrough makes no model call and changes no running system. It demonstrates the evidence, planning, confirmation, and verification boundaries Mqttable exposes through CLI and MCP.

QoS 1 retry case · 01

Start with evidence, not a guess.

The agent reads the relevant Trace rows, active Proxy toxics, and Monitor window. The scope stays tied to this broker and incident.

Trace · repeated PUBLISH, PUBACK outside timeout window
Proxy · latency_downstream active, 2400 ms ± 150 ms
Monitor · runtime healthy, no local pressure spike

Seven product surfaces, one investigation

Give the agent the same evidence a senior engineer would ask for.

Each page does useful work on its own. Together they preserve the context needed to diagnose, reproduce, change, and verify MQTT behavior.

01 · Connections

Start with live sessions, not screenshot guesses.

Connect MQTT 3.1.1 and MQTT 5 clients, publish, subscribe, inspect payloads, and keep protocol responses beside the client and broker that produced them.

Agent evidence: broker scope, client identity, ordered Trace rows, topics, QoS, payload metadata, and reason codes.

Mqttable Connections page showing a synthetic Factory Lab broker, MQTT client, and Trace rows in English.View full Connections screenshot

02 · Replay

Turn one intermittent failure into a repeatable scenario.

Create a scenario from Trace or PCAP evidence, review client mappings and timing, dry-run the plan, execute against a target broker, and inspect the run trace.

Agent evidence: source snapshot, executable steps, skipped rows, identity map, expected responses, run trace, and report history.

Mqttable Replay Workbench showing a synthetic scenario and reviewed dry run in English.View full Replay screenshot

03 · Bench

Keep the context that produced every throughput number.

Define client groups, subscriptions, scheduled messages, pacing, and run duration; then pair charts with preflight, host metrics, history, and comparison.

Agent evidence: exact load shape, preflight status, runtime context, message rates, latency, connection health, and retained reports.

Mqttable Bench page showing a synthetic point-to-point load scenario in English.View full Bench screenshot

04 · Proxy

Inject the failure before you trust the recovery.

Place a controlled proxy in front of the broker, add bounded latency, bandwidth, loss, timeout, or disconnect behavior, and verify cleanup afterward.

Agent evidence: listener and upstream scope, active toxics, direction, strength, schedule, Trace rows, cleanup state, and recovery check.

Mqttable Proxy page showing a synthetic downstream latency toxic in English.View full Proxy screenshot

05 · PCAP

Turn raw packets into root-cause evidence an agent can cite.

Analyze captures into connection context, issue groups, diagnostic checks, traffic summaries, packet rows, replay candidates, and bounded evidence references.

Agent evidence: source-backed findings, affected flows and clients, packet references, limitations, ranked causes, and next actions.

Mqttable PCAP page showing synthetic demo capture findings and packet-backed causes in English.View full PCAP screenshot

06 · Monitor

Decide whether the broker or local runtime is under pressure.

Read CPU, memory, file descriptors, scheduler pressure, process health, and structured product events before blaming the network or broker.

Agent evidence: bounded metric windows, runtime health, resource pressure, lifecycle facts, and correlated product events.

Mqttable Monitor page showing local runtime CPU, memory, resources, lifecycle, and trend charts in English.View full Monitor screenshot

07 · Tools

Keep sensitive debug material off unfamiliar websites.

Decode JWTs, compare payloads, hash files, inspect gzip data, convert timestamps, generate QR codes, and compress images inside the local workbench.

Local tools reduce unnecessary uploads of payloads, tokens, captures, certificates, and other debugging material.

Mqttable Tools page comparing two synthetic MQTT payloads in English.View full Tools screenshot

Simple Free / Pro boundary

Free for local MQTT debugging. Pro for scale and deeper evidence.

Free keeps daily MQTT work usable. Pro expands limits for serious Replay, PCAP, Benchmark, Remote Runtime, and advanced MCP-assisted diagnosis.

Free

$0/forever

For local MQTT debugging and daily trace work.

  • MQTT 3.1.1 and MQTT 5 clients with TCP, TLS, WebSocket, QoS, retain, will messages, publish, subscribe, trace, Monitor, Tools, and basic exports
  • Connections, PCAP Analysis, Replay, Proxy, Monitor, and Tools for everyday local diagnosis with practical Free limits
  • Benchmark runs within Free duration limits, including preflight, runtime context, and local evidence handoff
Early annual

Pro Annual

$49.90/year
$99

For engineers and consultants who need larger workspaces, deeper evidence, Remote Runtime, and Pro MCP workflows.

  • Larger Broker, Proxy, Replay, and PCAP workspaces plus Remote Runtime access
  • Replay above Free step limits, advanced pace controls, longer Bench duration, and complete PCAP diagnostic details
  • Advanced MCP workflows for deep diagnosis, TLS analysis, benchmark comparison, PCAP root-cause reports, and report persistence
  • Annual access to Pro MCP workflows, Remote Runtime, and MQTT evidence features
Lifetime

Pro Lifetime

$199

For engineers who want one-time Pro access for the supported life of Mqttable.

  • Larger Broker, Proxy, Replay, and PCAP workspaces plus Remote Runtime access
  • Replay above Free step limits, advanced pace controls, longer Bench duration, and complete PCAP diagnostic details
  • Advanced MCP workflows for deep diagnosis, TLS analysis, benchmark comparison, PCAP root-cause reports, and report persistence
  • One-time Pro access for the supported life of Mqttable, no renewal

Clear boundaries

AI-assisted does not mean unsupervised.

What Mqttable gives an agent, what stays local, and where a person remains in control.

Bring the evidence

Debug the next MQTT incident with the full story in view.

Download Mqttable, keep sensitive evidence local by default, and invite your agent into a bounded workflow when you need another pair of senior eyes.

Download Mqttable