Console Settings

Open Settings from the left navigation of the Device Agent console to manage shared model services, speech services, and platform connections. Platform deployments also show AI Credits and License.

Settings Entries

Manage an individual agent's prompt, timezone, vocabulary, voice tool scope, identity replies, and wait feedback through the agent card's More → Configuration overview. See Device Agent Configuration.

Entry	Contents
Platform	Platform provider and MQTT Config.
Models	Agent model, Vision model, Decision model, and Voice model.
Channels	Chat platform connections, enabled state, and allowed sources.
Permissions	Agent tool permissions and allowed command prefixes.
Webhook	Outbound notification connections for conversations and workflows.
AI Credits	Device consumption limits and balances.
License	Platform request authorization and validity.

Local runtimes can edit global settings. Daily editing options in platform deployments are:

Setting	Platform console action
Agent model, vision model, decision model, MQTT, and regular platform connection parameters	View settings; contact the deployment administrator for changes.
Voice, channels, and tool permissions	Change editable parameters and follow the status notice after saving.
Service credentials	View configuration status; contact the deployment administrator for replacement.
Webhook, AI Credits, and License	Manage through their entries; see the corresponding guides for requirements.

When a provider has a separate login entry, use it to authorize the connection.

Platform

Platform Provider

Open Platform → Platform provider to view the selected default, fleets, or thingsneural provider, its connection settings, model contract, topic templates, and payload examples. The provider determines device models and message protocols; use matching settings when connecting a device. See IoT Platform Integration for each provider’s setup.

With Default, Device Agent directly manages device models, device records, and reported state; devices connect through an SDK or MQTT. EMQX Fleets and ThingsNeural retain the management and protocols of their respective platforms.

See ThingsNeural platform example.

MQTT Configuration

Open Platform → MQTT Config to view the connection between Device Agent and the MQTT broker. Local runtimes can edit the following settings; platform services manage the connection and expose regular parameters as read-only.

Area	Settings	Notes
Endpoints	Broker URL, WebSocket URL, Client ID	brokerUrl is used by the gateway and device-side clients; wsUrl is used by the browser for MQTT over WebSocket.
Connection	MQTT version, Keep Alive, connect timeout, auto reconnect, reconnect period, Clean Start, session expiry	Session expiry is shown for MQTT 5.
Auth	Username, password	Matches the MQTT broker authentication settings.
TLS	TLS switch, certificate verification, CA, client certificate, client key, key passphrase, Server Name	Uploaded certificate files are stored under certificates in the runtime home.
Topic templates	Text requests, device commands, command responses, telemetry, events, NTP	Read-only in the current UI. Local custom topics use .env or the config file; platform topics follow the selected provider's protocol.

The local Gateway reconnects after saving in the console. Restart after changing .env or the config file directly. See Platform Operations to verify platform connections and restarts. For broker options and Zero EMQX setup, see MQTT Broker.

Models

Open Models → Agent model to select a provider and model, supply credentials, and adjust reasoning and iteration limits. Vision model controls image analysis: auto reuses an image-capable agent model; dashscope uses a separate model and API key. Local runtimes can save these settings; the platform console provides read-only access.

The model guide covers field definitions, reasoning levels, Cloudflare account and gateway IDs, Codex authentication, and startup examples. See Vision Configuration for image-analysis fields.

Decision Model

Open Models → Decision model to view or configure TypeSafe Jev's enabled state, model, API key, and timeout. See Jev Decision Model for setup, workflow decision nodes, and quick commands.

Voice Model

Open Models → Voice model, enable voice, select a region and provider, and configure recognition, synthesis, and credentials. The platform shows credential status; contact the deployment administrator to replace credentials. See Apply Behavior for saving and restart requirements.

For direct local connections, copy Voice connection URL and check that the client can reach its host and port. Platform clients use the public voice WebSocket URL published by the administrator. An HTTPS console requires wss://; see Bind Address and TLS for listener and public addresses.

See Voice Configuration for provider fields, ASR/TTS choices, vocabulary support, connection addresses, and .env examples. Configure each agent's recognition hotwords, voice tools, and identity replies, waiting messages, and timeout replies separately.

Channels

Open Channels to enable Feishu, DingTalk, Discord, Telegram, or Slack. Select the channel, check its credentials, and set the allowed sources. IM Access lists each platform's required credentials and setup steps.

allowFrom limits which users or conversation sources can access the device agent. Separate multiple values with commas or new lines. Local runtimes can edit credentials; the platform displays credential status and allows changes to editable fields such as enabled state, identifiers, and allowed sources. Restart a local runtime after saving to apply channel settings; on the platform, contact the deployment administrator to restart the relevant services after changing IM settings. Set an outbound notification destination in the request or workflow.

Permissions

The Permissions tab controls agent tool permissions. These settings restrict tool execution; they do not provide console user management or user role permissions.

Permission	Notes
tool_editor	Allows saving and deleting tool editor extensions.
write_file	Allows agents to write local files.
execute_command	Allows agents to run commands.
allowedPrefixes	Restricts command execution to allowed prefixes such as pwd, ls, or rg.

After saving, tool execution uses the current permission settings. Matching local environment variables override console values after a restart. See Environment Variables for startup options.

Webhook

The Webhook tab manages reusable outbound notification connections for Agent conversations and workflows.

You can save, test, enable, disable, edit, and delete connections without restarting Device Agent. A test sends a real message. A connection cannot be deleted while a workflow refers to it. See Create and Test a Connection for setup and presets, Configure a Custom Webhook for signing, templates, and response rules, and Delivery Boundaries for limits.

AI Credits

See AI Credits for per-device budgets, pricing rules, and balance adjustments.

License

See License for platform request limits in RPM, license validity, and updates.

Use Monitoring to verify runtime status after saving. See Platform Deployment for service setup and credential updates.