Agents require a paid Pinata plan. Upgrade here.
Get Your First Agent Running
This takes about 2 minutes. You’ll connect an LLM provider, create an agent, and start chatting.1. Add Your API Key
Your agent needs access to an LLM to think and respond. Head to agents.pinata.cloud/secrets and you’ll see cards for the major providers at the top. Click Connect on whichever you use — Anthropic, OpenAI, OpenRouter, Gemini, or Venice — paste your API key, and save. That’s it.2. Create Your Agent
Go to agents.pinata.cloud/agents and click Create Agent. You have two paths: Use a template if you want something working fast. Templates come pre-configured with skills, settings, and personality — just add your secrets and deploy. Browse templates at agents.pinata.cloud/templates. Build from scratch if you want full control. The wizard walks you through:- Identity — Pick a personality (Atlas, Nova, Sage) or create your own. Give it a name.
- Workspace — The default includes Node.js, Python, and common tools. Add skills if you want extra capabilities.
- Connect — Select your LLM provider and add any other API keys your agent needs.
- Deploy — Hit the button and wait about 30 seconds.
3. Start Talking
Once your agent is running, you’ll land on its chat interface. Say hello. Ask it to help you with something. It can write code, create files, search the web, and more.What’s in the Agent Dashboard
Click any agent to open its detail page. Here’s what you’ll find: Chat — Your conversation interface. You can switch models from the dropdown in the top right. Channels — Connect your agent to Telegram, Slack, or Discord so others can talk to it too. Snapshots — Version history of your agent’s workspace, plus git access for local editing. Skills — Packages that give your agent new abilities, like IPFS storage or on-chain registration. Secrets — API keys and credentials your agent can access as environment variables. Models — Configure which models your agent can use and set the default. Routes — Expose web servers or APIs running inside your agent to the internet. Tasks — Schedule prompts to run automatically.Dev Tools
When you need to debug or dig deeper:- Console — Full terminal access to your agent’s container
- Logs — Real-time logs with level filtering, search, and export
- Danger — Agent settings, OpenClaw version, restart, and delete
Working with Git
Want to edit your agent’s workspace locally? Clone it:Using the CLI
Everything in the UI is also available through the CLI:Next Steps
Templates
Deploy a pre-configured agent in one click
Channels
Let others talk to your agent on Telegram, Slack, or Discord
Skills
Give your agent new abilities
Routes
Expose a web app or API to the internet