Skip to main content
Version: v0.1 (Stable)

5-Minute Quickstart Tutorial

This hands-on tutorial walks you through creating your first autonomous AI agent, giving it system tools, and executing a multi-step task in ActonOS.

Step 1: Open the ActonOS Dashboard​

  1. Navigate to your running instance:
    • Local LAN: http://acton.local:8080 (or http://localhost:8080)
    • Remote: https://acton.your-tailnet.ts.net
  2. If prompted, enter the Admin PIN / Password you configured during onboarding.
  3. You will see the Operations & System Dashboard displaying live CPU, RAM, active agents, and event feeds.

Step 2: Open Agent Studio and Create an Agent​

  1. In the left navigation menu, click Agents (or press Ctrl + K and type Create Agent).
  2. Click the yellow + New Agent button in the top right.
  3. Fill in the Agent Identity:
    • Agent Name: Research Assistant
    • Avatar / Slug: research-bot
    • Description: Autonomous research agent with web browsing and document summarization tools.
  4. In the System Prompt section, define the persona: 
    You are an autonomous research assistant running inside ActonOS.
    Your mission is to perform accurate web research, synthesize data, and produce clean Markdown summaries.
    Always verify facts using available web search and scraping tools before giving answers.
  5. In the Model & LLM Cascade section:
    • Primary Model: Select anthropic:claude-3-7-sonnet or openai:gpt-4o.
    • Fallback Model: Select google:gemini-2.0-flash or ollama:llama3.
    • Temperature: 0.3 (for grounded, precise output).
  6. In the Tool Scopes & Permissions section:
    • Check native_web_search (DuckDuckGo / Tavily integration).
    • Check native_browser_headless (Chromium web scraper).
    • Check native_workspace_write (Save findings to workspace).
  7. Click Save Agent Manifest.

Step 3: Start a Chat & Test Tool Execution​

  1. In the navigation sidebar, click Chat.
  2. Select your newly created Research Assistant from the top agent dropdown.
  3. Type the following prompt into the message composer: 
    Please research the latest release notes for Go 1.26. Summarize the top 3 major language changes and save the summary to a file named 'go_126_summary.md' in the workspace.
  4. Click Send (or press Enter).

What Happens Behind the Scenes:​

  1. Model Cascade: ActonOS streams the prompt to your configured primary LLM via Server-Sent Events (SSE).
  2. Tool Invocations: The agent decides to call native_web_search("Go 1.26 release notes").
  3. Execution Card: A real-time collapsible card appears in the chat timeline showing the tool's input arguments, execution time, and raw JSON output.
  4. Autonomous File Write: The agent calls native_workspace_write to create /workspace/go_126_summary.md.
  5. Streaming Response: The agent provides a formatted Markdown summary directly in the chat window.

Step 4: Verify the Generated File in Workspace​

  1. In the sidebar, click Workspace.
  2. Notice the new file go_126_summary.md listed in the explorer tree.
  3. Click on the file to view it in the built-in CodeMirror Editor with live Markdown preview.
  4. Notice that ActonOS automatically indexed the file into Hybrid Semantic Memory (visible by the green Semantic Indexed badge).

What's Next?​

Congratulations! You have configured and executed your first autonomous agent in ActonOS.

Explore the rest of the guides to unlock advanced capabilities: