LightVela

What is LightVela

LightVela is a cloud-based AI Agent hosting platform launched by the Tencent Cloud team. LightVela deploys the open-source Hermes Agent deployment, operations and maintenance, and 24/7 operation are all hosted in the cloud. Users do not need servers or operations knowledge; they can create a dedicated AI assistant with just a few clicks on a web page and connect it to WeChat, QQ, WeCom, Feishu, and DingTalk conversations. LightVela supports multiple models, skill extensions, persona injection, scheduled tasks, and cross-session memory, and can host multiple Agents.

Main features of LightVela

• Cloud hosting: The Agent's deployment, operations and maintenance, and 24/7 operation are all completed in the cloud, with no need for servers or operations knowledge, and can be configured with just a few clicks on a web page.
• Multi-channel access: Supports the five major IM platforms WeChat, QQ, WeCom, Feishu, and DingTalk; after scanning a QR code to authorize, you can chat directly with the Agent in the messaging app.
• Multi-model support: Can connect to mainstream domestic models such as DeepSeek, Kimi, Doubao, MiniMax, and Qwen, and supports three methods: Coding Plan monthly subscription, API pay-as-you-go, and OpenAI-compatible endpoints.
• Skill extensions: Install individual skills or skill packs through Tencent SkillHub, and the Agent automatically determines when to invoke them on demand.
• Persona injection: Supports three methods: random blind-box draws, testing personas, and uploading persona files, giving the Agent a unique personality and way of expressing itself.
• Scheduled tasks: Supports AI intelligent tasks as well as notification tasks and script tasks that do not consume credits, and can set upstream task associations to achieve chained task execution.
• Memory accumulation: Automatically accumulates Agent memory and user profiles, persists across sessions, and understands you better the more you use it.
• Multi-Agent hosting: Can create multiple Agents, each with an independent cloud environment, with channels, models, skills, personas, and memory configured separately and isolated from one another.

How to use LightVela

• Register and log in: Visit the LightVela official website https://lightvela.com/ and complete account login and registration.
• Create an Agent: Choose an avatar, name it, fill in a description, and create your dedicated Hermes Agent with one click.
• Configure models: Experience credits are preset by default and ready to use out of the box. In Configure Models, you can connect your own API Key, Coding Plan, or OpenAI-compatible endpoint.
• Connect channels: In Configure Channels, scan a QR code to authorize any IM among WeChat, QQ, WeCom, Feishu, or DingTalk, or use it directly on the web page.
• Install skills: Search and select the required skills or skill packs in SkillHub, and the Agent will automatically invoke them based on the conversation context.
• Inject a persona: Set the Agent's personality and expression style by randomly drawing a blind box, testing, or uploading a persona file.
• Set scheduled tasks: Choose a template to configure AI intelligent, notification, or script tasks, and associate upstream tasks to form a task chain.
• Start chatting: Chat with the Agent directly in IM or on the web page to assign tasks, and the cloud automatically runs them and returns the results.

LightVela's core advantages

• 24/7 cloud residency: The Agent runs online around the clock; after tasks are issued, there is no need for a local device to be powered on, completely eliminating dependence on desktop clients.
• Native integration with Tencent IM platforms: WeChat, QQ, WeCom, Feishu, and DingTalk can all be connected by scanning a QR code; this entry-point advantage is hard for other competitors to replicate.
• Cross-session long-term memory: Automatically accumulates MEMORY.md and user profiles, so the Agent understands you better the more you use it, instead of reintroducing yourself every time.
• Independent hosting for multiple Agents: You can raise multiple "digital employees", with resource isolation and independent configuration, meeting the needs of division of labor and collaboration.
• Freedom to choose models: Compatible with mainstream domestic models and any OpenAI endpoint, so you are not locked in by a single model vendor.
• SkillHub skill ecosystem: Backed by Tencent's skill library, skill packages can be installed with one click, and capability boundaries can be expanded infinitely.

Comparison of LightVela with similar competitors

LightVela application scenarios

• Personal knowledge management and topic selection assistant: Let the Agent continuously search the internet for the latest content in fields of interest, automatically save it to the knowledge base and organize and summarize it, assisting with content topic selection and decision-making.
• Self-media/WeChat official account operations: Configure writing and formatting skill packages; once a topic is given, the Agent can generate article outlines, drafts, and even formatting materials in the cloud, and with scheduled tasks can automate the daily morning/evening report output.
• Development companion and automated operations: For developers, PRD writing and code assistance skills can be configured, and script tasks that do not consume credits can be used to regularly execute inspection-type and deterministic automation work.
• Finance/business assistant: After installing finance-related skills, it periodically fetches market and financial report data and generates analysis reports, runs 7×24 in the cloud, and proactively pushes reminders to WeChat when trigger conditions are met.
• Life Assistant: Set up notification-type scheduled tasks, such as reminders at fixed times to drink water, rest, or for schedule items; combined with persona injection and long-term memory, the Agent becomes increasingly handy the more you use it, in your preferred communication style.