Skip to main content
Mqttable
Sign In

Open navigation

Mqttable

Mqttable for AI Agents

Give AI agents MQTT evidence, not uncontrolled access.

Mqttable lets compatible agents discover the runtime, read bounded MQTT evidence, and execute authorized operations through one mutation entry point. The server handles commit checks and result evidence.

DownloadAgent quickstart

Synthetic walkthrough: static text, not live telemetry. Mqttable does not call or bundle an independent model.

mqttable · mcp
Static transcript

01 · Discover

$ query_mqttable → mqttable://runtime/status find_mqttable_capabilities → describe_mqttable_capability

Runtime is loopback, MCP is enabled, runtime health is ready, and the schema includes its compact example.

02 · Capture

$ query_mqttable → mqtt.qos.analyze_flow query_mqttable → proxy.toxic.list

QoS analysis finds repeated PUBLISH packets and a PUBACK at 2.6 s; Proxy reports 2400 ms downstream latency.

03 · Reproduce

$ query_mqttable → bounded Trace URI query_mqttable → proxy.toxic.list

Bounded Trace, Proxy, and Monitor evidence align on downstream latency while local runtime pressure stays normal.

04 · Act

$ execute_mqttable_action({ action: "proxy.toxic.delete", input: { … }, idempotency_key: "remove-fault-once" })

One operation targets only latency_downstream. A stable idempotency key prevents request retries from deleting or executing twice.

05 · EVIDENCE

$ query_mqttable → operation.read query_mqttable → mqtt.qos.analyze_flow

Use returned operation evidence. Query operation.read only when pending or uncertain, then inspect MQTT flow evidence for the user’s goal.

Fixed QoS 1 latency walkthrough. It does not auto-play, call a model, connect to a broker, or represent live runtime output.

QoS 1 latency retry

Discover, capture, reproduce, act, and verify against one bounded incident.

Read tools have no business side effects. A single execute call checks, performs and observes a supported change while MCP is enabled and runtime health is ready.

  1. 01

    Discover

    Runtime is loopback, MCP is enabled, runtime health is ready, and the schema includes its compact example.

  2. 02

    Capture

    QoS analysis finds repeated PUBLISH packets and a PUBACK at 2.6 s; Proxy reports 2400 ms downstream latency.

  3. 03

    Reproduce

    Bounded Trace, Proxy, and Monitor evidence align on downstream latency while local runtime pressure stays normal.

  4. 04

    Act

    One operation targets only latency_downstream. A stable idempotency key prevents request retries from deleting or executing twice.

  5. 05

    EVIDENCE

    Use returned operation evidence. Query operation.read only when pending or uncertain, then inspect MQTT flow evidence for the user’s goal.

Compact external contract

Five tools keep discovery narrow and execution direct.

Examples are included in schema expansion. Capability depth can change without teaching agents a new top-level tool list.

find_mqttable_capabilities

Finds one typed action, workflow, resource, or prompt when its capability id is unknown.

describe_mqttable_capability

Returns one exact action, resource, or workflow contract with a compact example.

query_mqttable

Queries data, runs no-effect analysis, or reads one bounded mqttable:// resource.

execute_mqttable_action

Executes one authorized action with effects and returns durable operation evidence.

diagnose_mqttable

Returns passive or explicitly active diagnostics for the Mqttable runtime and MCP service.

Persistent local policy

One switch, separate runtime health

Enable MCP to authorize supported reads and changes. Runtime checks still decide whether a change can execute.

mcp_enabled: false

MCP unavailable

The endpoint is off. URL-only client configs work again after the user re-enables access.

mcp_enabled: true

MCP enabled

Local clients can read evidence and execute supported operations. Enablement persists until you turn MCP off.

Runtime health is separate from enablement. If a change service is unavailable, writes are rejected without changing your setting; available diagnostic reads remain accessible.

Hard boundaries

Agent access stays narrower than the workbench.

Writes stay on the local machine

Local MCP clients connect to the loopback endpoint without a client credential. Browser Origin requests are rejected, and mutable calls still require the live change gate.

Remote MCP is not included

MCP accepts loopback HTTP only. Non-loopback requests have no configuration or authorization-provider bypass.

Raw secrets stay outside MCP

Passwords, bearer tokens, PEM certificates, and private keys are entered through first-party intake. Agent flows reference SecretRef, CertRef, and KeyRef handles.

Replay stays in Workbench

Replay scenarios are created, reviewed, dry-run, and executed in the Workbench. There is no current Replay-specific MCP execution surface.

Three-step setup

Connect an agent without inventing configuration.

Read the full agent quickstart
  1. 01

    Open Settings → MCP and enable MCP for trusted local clients. Changes still require current authorization and healthy runtime gates.

  2. 02

    Copy the generated URL-only client configuration. MCP off/on keeps it valid without a client credential lifecycle.

  3. 03

    Read runtime info first, search only when the capability id is unknown, then read its schema before calling it.

Agent FAQ

The useful questions before connecting an agent.

Give your agent bounded MQTT evidence without giving up control.

Open the Download page, choose the MCP access mode in Settings, and start with runtime discovery rather than guessed state.

Download