Tuya releases Muse Skill for smart devices across 3,000+ categories
The open-source Tuya Cloud Devices Skill lets Meta’s Muse query and control devices through natural-language requests, using Tuya’s cloud rather than local device keys or firmware changes. Its functions include device control and status checks, with specific limits and extra requirements for event-based features.
Tuya says its platform spans more than 3,000 smart hardware categories.1
Tuya reports that its platform serves more than 200 countries and regions.1
By June 30, 2026, Tuya reported more than 2.09 million registered AI developers.1
Story
How Tuya Cloud Devices brings smart-home control to MuseTuya Smart has released Tuya Cloud Devices, an official open-source Skill that enables Meta’s Muse to ask about and operate smart devices through natural-language requests.1 The announcement places the Skill within a platform that spans 3,000+ smart hardware categories and is used across 200+ countries and regions.1 Those figures describe the range and geographic reach of Tuya’s platform, not a count of devices already connected to Muse through this release.1 Tuya says the Skill uses its cloud service, so setup does not call for local device keys, a shared local network, or device firmware alterations.1 The German-language announcement likewise described access to devices in a Tuya account with natural-language requests and one API key, without local keys or firmware changes.1
The release addresses a more involved approach used by community methods for controlling devices that were already installed in homes.1 Those methods generally asked users to retrieve a local key for every device and check its protocol version and data-point definitions.1 If a device was paired to the network again, its earlier key became invalid, so the user had to go through that work again.1 Tuya’s announcement describes its cloud-based method as requiring neither those local keys nor firmware changes, while the German account also refers to one API key.1 This is a change in the stated access method; the announcement does not give a measured reduction in setup time or a count of users who have switched.1
Muse began operating in September 2026 and runs on a cloud virtual machine assigned to each user.1 It can keep tasks such as handling email and scheduling active in the background even when its application has been closed.1 Meta has also made Muse Gadgets open source, providing a hardware interface for developers to create add-ons.1 The described development options include Wi-Fi MCU boards and Raspberry Pi, used with Meta’s open-source SDK.1 Tuya Cloud Devices was first made for OpenClaw and was adapted to work with Muse by following the Muse Gadgets specification.1
The Skill’s name in the Muse environment is gadget-tuya-cloud-devices, and Tuya identifies it as an official Skill derived from its open-source tuya-smart-control AI skill.1 Its implementation is a Markdown file that follows the Muse Gadgets specification.1 Tuya engineers adapted that file and released it as open source.1 In the arrangement described by Tuya, Muse contributes the conversational and agent functions, while Tuya provides the connection to devices.1 The release therefore describes a software integration between Muse and Tuya’s device services, rather than a change to the devices’ firmware.1
Users can apply for a free API key through Tuya’s AI website, tuya.ai.1 The key starts with the characters “sk-”; the following two characters indicate the account’s data center.1 The Skill uses that identifier to choose an API endpoint from seven data centers without requiring a user to set the endpoint by hand.1 The key is included in request headers, is not repeated in conversation, and is limited to devices belonging to the user’s own account.1 The stated account restriction defines the scope of access for the key; the announcement does not claim that it grants access to other users’ devices.1
When a user requests an action, the Skill first checks whether a device-specific or local-control Skill is available and routes the request there first if one is.1 For a device it handles, the Skill consults that device’s standardized Thing Model before creating a command.1 The model provides information about permitted operations, acceptable values, and the increments by which values can change.1 Muse can operate devices, report their status, organize homes and rooms, and respond to practical questions using Tuya devices.1 Together, the routing rule and model check describe how the Skill chooses where to handle a request and what device constraints apply.1
When a request could refer to several devices, the Skill presents their room locations and asks the user to choose before proceeding.1 If a device is offline, the Skill reports that condition and does not send it a command.1 After issuing an instruction, it checks the device’s status one to two seconds later and confirms success only if the check shows that the requested change occurred.1 Users may receive alerts by SMS, voice call, email, or app push notification.1 Device-event notifications are spaced at least 30 minutes apart, a limit Tuya says is intended to reduce alert fatigue.1
A camera-related request can produce a still image or a video lasting up to 60 seconds.1 Real-time events and notifications triggered by events require an always-online Linux device to receive subscriptions to those events.1 That Linux requirement applies to the event and notification functions described in the release, while the general cloud-based operation is described separately.1 The Skill does not provide door-lock operation, live video, firmware updates, network setup, or device removal.1 For those excluded actions, it sends the user to the Tuya app instead.1
Tuya reports that, as of June 30, 2026, its registered AI developer total exceeded 2.09 million.1 That figure counts registered AI developers on the Tuya platform; it is distinct from the platform’s 3,000+ hardware categories and its reach across 200+ countries and regions.1 The announcement gives those platform measures as context for a Skill focused on connecting Muse to Tuya devices, but does not say how many people will use the Skill.1 Its stated operating model combines Muse’s conversational and agent capabilities with Tuya’s connectivity, while device access remains bounded by the user’s account and the listed exclusions.1 The release thus sets out both the available controls and the conditions attached to them, including device-status checks and the additional Linux requirement for event subscriptions.1
History
How it came to this- June 30, 2026Tuya reports its AI developer countThe company reported more than 2.09 million registered AI developers on its platform.
- September 2026Muse launchesMuse launched as a cloud-based service with a virtual machine dedicated to each user.
- Now3,000+ Tuya platform smart hardware categories
Impact
Spreading outward, level by level- Level 1Device access
The Skill lets Muse query and control Tuya-connected smart devices in natural language through Tuya’s cloud, without local device keys, a shared local network, or firmware changes.1
Fact - Level 2Setup compared with community methods
Earlier community approaches involved retrieving each device’s local key and checking its protocol and data-point definitions; re-pairing a device invalidated its key and meant repeating the process. The announced cloud method avoids local device keys.1
Fact - Level 3Request handling and safeguards
The Skill checks device models before generating commands, asks users to clarify when multiple devices match, and reports offline devices without sending commands. It checks device status after a command before confirming success.1
Fact - Level 4Event features and exclusions
Event subscriptions require an always-online Linux device. The Skill also excludes several operations, including door-lock control and live video, directing those requests to the Tuya app.1
Fact - Level 5Platform context
The category and geographic figures describe Tuya’s platform reach; they do not establish how many devices or users will connect to Muse through this Skill.
Analysis
Sources
What each source supportsWritten by AI from the sources listed below: every fact was checked word for word against its source, and inference is marked apart. How we write