# Table of Contents - [Documentation - Agent Zero](#documentation-agent-zero) - [Sharing Agent Zero Instances - Agent Zero](#sharing-agent-zero-instances-agent-zero) - [Tasks & Scheduling - Agent Zero](#tasks-scheduling-agent-zero) - [MCP & A2A Integration - Agent Zero](#mcp-a2a-integration-agent-zero) - [Multi-Agent Cooperation - Agent Zero](#multi-agent-cooperation-agent-zero) - [Secrets & Variables - Agent Zero](#secrets-variables-agent-zero) - [Community - Agent Zero](#community-agent-zero) - [Working with Projects - Agent Zero](#working-with-projects-agent-zero) - [Updating Agent Zero - Agent Zero](#updating-agent-zero-agent-zero) - [Memory Management - Agent Zero](#memory-management-agent-zero) - [Skills - Agent Zero](#skills-agent-zero) - [Get Started - Agent Zero](#get-started-agent-zero) - [Prompts & Customization - Agent Zero](#prompts-customization-agent-zero) - [Changelog - v0.6 - Agent Zero](#changelog-v0-6-agent-zero) - [Extensions Framework - Agent Zero](#extensions-framework-agent-zero) - [Changelog - v0.8.5 - Agent Zero](#changelog-v0-8-5-agent-zero) - [Changelog - v0.8.1 - Agent Zero](#changelog-v0-8-1-agent-zero) - [Changelog - v0.7.1 - Agent Zero](#changelog-v0-7-1-agent-zero) - [Changelog - v0.9.5.1 - Agent Zero](#changelog-v0-9-5-1-agent-zero) - [Changelog - v0.8.2 - Agent Zero](#changelog-v0-8-2-agent-zero) - [Changelog - v0.9.0 - Agent Zero](#changelog-v0-9-0-agent-zero) - [Changelog - v0.9.4 - Agent Zero](#changelog-v0-9-4-agent-zero) - [Changelog - v0.9.7 - Agent Zero](#changelog-v0-9-7-agent-zero) - [Changelog - v0.9.1 - Agent Zero](#changelog-v0-9-1-agent-zero) - [Changelog - v0.9.1.1 - Agent Zero](#changelog-v0-9-1-1-agent-zero) - [Changelog - v0.8.2.1 - Agent Zero](#changelog-v0-8-2-1-agent-zero) - [Changelog - v0.6.5 - Agent Zero](#changelog-v0-6-5-agent-zero) - [Changelog - v0.6.1 - Agent Zero](#changelog-v0-6-1-agent-zero) - [Changelog - v0.3 - Agent Zero](#changelog-v0-3-agent-zero) - [Changelog - v0.9.5 - Agent Zero](#changelog-v0-9-5-agent-zero) - [Changelog - v0.5 - Agent Zero](#changelog-v0-5-agent-zero) - [Changelog - v0.8.1.2 - Agent Zero](#changelog-v0-8-1-2-agent-zero) - [Changelog - v0.8.3.1 - Agent Zero](#changelog-v0-8-3-1-agent-zero) - [Changelog - v0.9.3 - Agent Zero](#changelog-v0-9-3-agent-zero) - [Changelog - v0.8.6 - Agent Zero](#changelog-v0-8-6-agent-zero) - [Changelog - v0.2 - Agent Zero](#changelog-v0-2-agent-zero) - [Changelog - v0.8.4.1 - Agent Zero](#changelog-v0-8-4-1-agent-zero) - [Changelog - v0.8.7 - Agent Zero](#changelog-v0-8-7-agent-zero) - [Changelog - v0.6.4 - Agent Zero](#changelog-v0-6-4-agent-zero) - [Changelog - v0.9.6 - Agent Zero](#changelog-v0-9-6-agent-zero) - [Changelog - v0.7 - Agent Zero](#changelog-v0-7-agent-zero) - [Changelog - v0.6.3 - Agent Zero](#changelog-v0-6-3-agent-zero) - [Changelog - v0.8.4.2 - Agent Zero](#changelog-v0-8-4-2-agent-zero) - [Changelog - v0.8.1.1 - Agent Zero](#changelog-v0-8-1-1-agent-zero) - [Changelog - v0.8.4 - Agent Zero](#changelog-v0-8-4-agent-zero) - [Changelog - v0.8.3 - Agent Zero](#changelog-v0-8-3-agent-zero) - [Changelog - v0.6.2 - Agent Zero](#changelog-v0-6-2-agent-zero) - [Changelog - v0.4 - Agent Zero](#changelog-v0-4-agent-zero) - [Changelog - v0.8 - Agent Zero](#changelog-v0-8-agent-zero) - [Changelog - v0.9.2 - Agent Zero](#changelog-v0-9-2-agent-zero) - [Changelog - v0.8.5.1 - Agent Zero](#changelog-v0-8-5-1-agent-zero) - [Changelog - v0.9.8 - Agent Zero](#changelog-v0-9-8-agent-zero) - [Changelog - Agent Zero](#changelog-agent-zero) --- # Documentation - Agent Zero [Home](https://www.agent-zero.ai/) › Docs ![Agent Zero logo](https://www.agent-zero.ai/res/lightSymbol.svg) ![Agent Zero logo](https://www.agent-zero.ai/res/darkSymbol.svg) Documentation ============= Explore guides, capabilities, and resources for building, extending, and operating Agent Zero. ### [Get Started](https://www.agent-zero.ai/p/docs/get-started/) Install Agent Zero, configure providers, and launch your first instance. [](https://www.agent-zero.ai/p/docs/get-started/) ### [Updating](https://www.agent-zero.ai/p/docs/updating/) Safe upgrade paths, backups, and how to verify new releases. [](https://www.agent-zero.ai/p/docs/updating/) ### [Projects](https://www.agent-zero.ai/p/docs/projects/) Organize work with per-project instructions, memory, knowledge, and files. [](https://www.agent-zero.ai/p/docs/projects/) ### [Skills](https://www.agent-zero.ai/p/docs/skills/) Create contextual expertise modules with the cross-platform SKILL.md standard. [](https://www.agent-zero.ai/p/docs/skills/) ### [MCP / A2A](https://www.agent-zero.ai/p/docs/mcp-a2a/) Connect via Model Context Protocol and Agent-to-Agent flows for interoperability. [](https://www.agent-zero.ai/p/docs/mcp-a2a/) ### [Subagents](https://www.agent-zero.ai/p/docs/subagents/) Create specialized subagents with dedicated tools, prompts, and roles. [](https://www.agent-zero.ai/p/docs/subagents/) ### [Memory](https://www.agent-zero.ai/p/docs/memory/) Store, retrieve, and manage agent memories through the Memory Dashboard. [](https://www.agent-zero.ai/p/docs/memory/) ### [Task Scheduler](https://www.agent-zero.ai/p/docs/task-scheduler/) Schedule recurring or one-off jobs so Agent Zero runs tasks on time. [](https://www.agent-zero.ai/p/docs/task-scheduler/) ### [Secrets](https://www.agent-zero.ai/p/docs/secrets/) Manage credentials securely so agents can use them without exposure. [](https://www.agent-zero.ai/p/docs/secrets/) ### [Sharing](https://www.agent-zero.ai/p/docs/sharing/) Share access safely with teammates using tunnels, auth, and best practices. [](https://www.agent-zero.ai/p/docs/sharing/) ### [Prompts](https://www.agent-zero.ai/p/docs/prompts/) Author, version, and reuse prompts that shape Agent Zero's behavior. [](https://www.agent-zero.ai/p/docs/prompts/) ### [Extensions](https://www.agent-zero.ai/p/docs/extensions/) Discover and integrate extensions to expand capabilities without forking. [](https://www.agent-zero.ai/p/docs/extensions/) ### [Community](https://www.agent-zero.ai/p/docs/community/) Join Discord and Skool to get help, share builds, and stay up to date. [](https://www.agent-zero.ai/p/docs/community/) ### [Changelog](https://www.agent-zero.ai/p/docs/changelog/) Track every release, feature, and fix across Agent Zero versions. [](https://www.agent-zero.ai/p/docs/changelog/) We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Sharing Agent Zero Instances - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Sharing Access & Security ================= Sharing Agent Zero Instances ---------------------------- Share your Agent Zero instance securely with teammates and collaborators using authentication and network access. For complete instructions on network access and secure sharing, see the **Mobile & Network Access** section in the Get Started guide. [→ View Network Access Guide](https://www.agent-zero.ai/p/docs/get-started/#network-access) ### Quick Start: Share via Tunnel 1. **Enable Authentication:** Settings → Authentication → Set username and password 2. **Open External Services:** Settings → External Services 3. **Create Tunnel:** Scroll to Flare Tunnel section → Click "Create Tunnel" 4. **Share Access:** Scan QR code or share the provided URL **⚠️ Security Warning:** Always enable authentication (UI login/password) before exposing to the internet! ### Local Network Sharing Share on your local network (home/office): 1. Find your computer's local IP address 2. Share with colleagues: `http://[YOUR_IP]:[PORT]` 3. Example: `http://192.168.1.100:50080` ### Additional Resources * [Authentication Setup](https://www.agent-zero.ai/p/docs/get-started/#configuration) * [Complete Network Access Guide](https://www.agent-zero.ai/p/docs/get-started/#network-access) * [Projects for Multi-User Isolation](https://www.agent-zero.ai/p/docs/projects/) We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Tasks & Scheduling - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Task Scheduler Automation ========== Tasks & Scheduling ------------------ Automate recurring workflows with Agent Zero's powerful task scheduler. Create scheduled cron jobs, plan specific executions, and integrate with projects for context-aware automation. What are Tasks? --------------- Agent Zero includes a built-in task scheduler that enables unattended automation. Tasks are agent executions that run automatically based on schedules, specific dates, or manual triggers - perfect for recurring workflows like reports, monitoring, and data processing. ### Task Types * **Scheduled Tasks:** Recurring executions using cron syntax (daily reports, hourly checks) * **Planned Tasks:** Specific date/time list (one-time deployments, timed notifications) * **Ad-hoc Tasks:** Manual execution only (on-demand operations, testing) **💡 Integration:** Tasks work exceptionally well with Projects, allowing automated workflows with project-specific context, secrets, and memory. Accessing the Scheduler ----------------------- ### Via Dashboard 1. Click **Dashboard** in the sidebar 2. Click the **Scheduler** button 3. The Task Scheduler modal opens, showing all tasks ![Task Scheduler Interface](https://www.agent-zero.ai/res/docs/usage/tasks/scheduler-1.png) ### What You'll See * **Task List:** All tasks with status indicators * **Task Details:** Name, project association, last run, next run time * **Quick Actions:** Edit, enable/disable, run now, delete Creating Tasks -------------- ### Via User Interface 1. Open the Task Scheduler from the Dashboard 2. Click **"Create Task"** 3. Configure task settings (see Configuration section below) 4. Click **"Save"** ### Via Chat Ask the agent to create tasks programmatically: **Example:** Create a scheduled task that checks my email inbox every morning at 8 AM and summarizes new messages from the last 24 hours. The agent uses the `scheduler` tool to create tasks based on your natural language description. Task Configuration ------------------ ![Edit Task Interface](https://www.agent-zero.ai/res/docs/usage/tasks/edit-task.png) ### Task Name A unique identifier for the task: * Use descriptive names: "Morning report for incoming Gmail" * Keep it concise but clear * Helps identify tasks in logs and notifications ### Task Type Choose the execution model: * **Scheduled:** Recurring cron-based execution * **Planned:** Specific date/time list * **Ad-hoc:** Manual execution only **📝 Note:** Task type cannot be changed after creation. To switch types, create a new task. ### Project Association Optionally link the task to a project: #### Benefits of project-scoped tasks: * Task inherits project instructions automatically * Access to project-specific secrets and variables * Uses project's isolated memory (if configured) * Runs in project's working directory * Perfect for client-specific or domain-specific automation **Example:** * **Project:** "Financial Reports" * **Task:** "Daily Portfolio Summary" * **Result:** Task automatically uses financial analysis instructions, API keys, and memory from the project ### State Management Control task execution: * **Idle:** Ready to run when scheduled (green indicator) * **Running:** Currently executing (blue indicator) * **Disabled:** Won't execute even if scheduled (gray indicator) * **Error:** Last execution failed (red indicator) Toggle state to temporarily disable tasks without deleting them. ### Execution Plan #### For Scheduled Tasks (Cron Syntax) Configure recurring execution using cron patterns: # Every day at 9 AM 0 9 * * * # Every hour 0 * * * * # Every Monday at 10 AM 0 10 * * 1 # Every 15 minutes */15 * * * * # First day of month at midnight 0 0 1 * * **Timezone aware:** Tasks respect your configured timezone setting #### For Planned Tasks Add specific execution times: 1. Click "Add Execution Time" 2. Pick date and time 3. Tasks execute in order 4. Completed executions move to "Done" list Task Execution -------------- ### Dedicated vs. Shared Context #### Dedicated Context (recommended): * Each task has its own isolated chat history * Prevents context pollution between tasks * Better for independent, recurring workflows * Default for UI-created tasks #### Shared Context: * Task shares a chat context * Can build on previous task executions * Useful for sequential, related operations * Typically used for agent-created tasks ### Execution Flow When a task runs: #### 1\. Context Preparation * Creates or loads the task's chat context * Activates associated project (if configured) * Loads project instructions, secrets, and memory #### 2\. Prompt Injection * Injects system prompt (if specified) * Adds task-specific instructions * Includes any file attachments #### 3\. Agent Execution * Agent receives the prompt as a user message * Executes using available tools * Can use project-scoped resources #### 4\. Result Handling * Stores execution result and timestamp * Updates task state (idle or error) * Sends notifications (if configured) * For Planned Tasks: moves execution to "Done" list ### Monitoring Execution #### Real-time monitoring: * Running tasks show "Running" state in scheduler * View task's chat context to see agent progress * Stop tasks mid-execution if needed #### Execution history: * **Last Run:** Timestamp of most recent execution * **Next Run:** Scheduled time for next execution * **Last Result:** Output or error from last run Common Use Cases ---------------- ### Daily Reports Task: "Morning Inbox Summary" Type: Scheduled (0 9 * * *) Project: "Email Automation" Prompt: "Check my Gmail inbox for new messages from the last 24 hours. Summarize important emails by category and highlight any urgent items." ### Periodic Monitoring Task: "Server Health Check" Type: Scheduled (*/30 * * * *) Project: "Infrastructure" Prompt: "Check all production servers. Report any issues with CPU usage over 80%, memory over 90%, or disk space below 10%." ### Data Processing Task: "Daily Sales Analysis" Type: Scheduled (0 1 * * *) Project: "Sales Analytics" Prompt: "Process yesterday's sales data from /data/sales/. Generate summary report with trends, top products, and anomalies. Save report to /reports/ and email to team." ### Scheduled Deployments Task: "Weekly Production Deploy" Type: Planned (Sundays at 2 AM) Project: "DevOps" Prompt: "Deploy latest release to production environment. Run post-deployment tests. Alert on-call engineer if issues detected." ### Content Publishing Task: "Social Media Updates" Type: Scheduled (0 10,14,18 * * *) Project: "Marketing" Prompt: "Check /content/queue/ for scheduled posts. Publish the next item to Twitter, LinkedIn, and Facebook. Move published content to /content/archive/." Best Practices -------------- ### Task Design * **Be specific:** Clear, detailed prompts lead to consistent results * **Include error handling:** Tell the agent what to do if something fails * **Define success criteria:** How should the agent verify completion? * **Use projects:** Leverage project-scoped context and secrets ### Scheduling * **Consider dependencies:** Stagger tasks that depend on each other * **Avoid peak hours:** Schedule heavy tasks during low-usage periods * **Test timing:** Ensure tasks complete before the next execution * **Use appropriate frequency:** Don't over-schedule; balance freshness vs. resources ### Monitoring * **Check execution logs:** Review task results regularly * **Set up notifications:** Alert on failures or important events * **Test tasks manually:** Use ad-hoc execution to verify behavior * **Keep prompts updated:** Adjust as requirements or APIs change Need Help? ---------- Questions about task scheduling? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # MCP & A2A Integration - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Mcp A2a Integration =========== MCP & A2A Integration --------------------- Extend Agent Zero with external MCP servers for additional tools and enable Agent-to-Agent communication for distributed workflows. MCP Server Setup ---------------- Agent Zero can connect to external MCP (Model Context Protocol) servers to extend its capabilities with additional tools. ### What are MCP Servers? MCP servers are external tools that Agent Zero can use to perform specialized tasks. Popular examples include: * **Browser automation** - Chrome DevTools, Playwright * **Workflow automation** - n8n integration * **Email operations** - Gmail management * **Database access** - SQLite, PostgreSQL * **File operations** - Advanced file system tools * **API integrations** - Third-party service connectors **📝 Note:** This guide covers connecting to external MCP servers as a **client**. For exposing Agent Zero as an MCP **server**, see the advanced connectivity documentation. ### Step 1: Open MCP Configuration 1. Click **Settings** in the sidebar 2. Navigate to the **MCP/A2A** tab 3. Click on **External MCP Servers** 4. Click the **Open** button to access the configuration editor ![MCP Configuration Access](https://www.agent-zero.ai/res/docs/setup/mcp/mcp-open-config.png) ### Step 2: Add Your MCP Server In the JSON editor, add your MCP server configuration. Here's a simple example: { "mcpServers": { "chrome-devtools": { "command": "npx", "args": ["-y", "chrome-devtools-mcp@latest"] } } } ![MCP Configuration Example](https://www.agent-zero.ai/res/docs/setup/mcp/mcp-example-config.png) ### Step 3: Apply and Verify 1. Click **Apply now** to save your configuration 2. The server status will appear below, showing: * Server name (e.g., `chrome_devtools`) * Number of available tools * Connection status (green indicator = connected) **💡 Tip:** The first time you run an `npx`\-based MCP server, it may take a few moments to download and initialize. ### Common MCP Server Examples #### Local Command-Based Server { "mcpServers": { "sqlite": { "command": "uvx", "args": ["mcp-server-sqlite", "--db-path", "/root/db.sqlite"] } } } #### Remote HTTP Server { "mcpServers": { "external-api": { "url": "https://api.example.com/mcp", "headers": { "Authorization": "Bearer YOUR_API_KEY" } } } } #### Multiple Servers Configuration { "mcpServers": { "browser": { "command": "npx", "args": ["-y", "chrome-devtools-mcp@latest"] }, "database": { "command": "uvx", "args": ["mcp-server-sqlite", "--db-path", "/data/app.db"] }, "external-api": { "url": "https://api.example.com/mcp", "headers": { "Authorization": "Bearer token123" } } } } ### Docker Networking If Agent Zero runs in Docker and your MCP server runs on the host: * **macOS/Windows:** Use `host.docker.internal` in URLs { "url": "http://host.docker.internal:PORT/endpoint" } * **Linux:** Run the MCP server in the same Docker network and use the container name { "url": "http://container_name:PORT/endpoint" } ### Using MCP Tools Once connected, MCP tools become available to Agent Zero automatically. Tools are named with the server prefix: **Example:** * Server name: `chrome-devtools` * Tool becomes: `chrome_devtools.navigate_to_url` Simply ask Agent Zero to perform tasks, and it will use the appropriate MCP tools when needed. ### Recommended MCP Servers Community-tested and reliable MCP servers: * **Chrome DevTools MCP** - Direct Chrome control * **Playwright MCP** - Cross-browser automation * **n8n MCP** - Workflow automation * **Gmail MCP** - Email management * **VSCode MCP** - IDE workflows **💡 Tip:** For browser automation tasks, MCP-based browser tools are more reliable than the built-in browser agent. A2A Server Setup ---------------- Agent Zero can communicate with other Agent Zero instances using the A2A (Agent-to-Agent) protocol based on FastA2A. ### What is A2A? A2A enables direct communication between multiple Agent Zero instances. This allows: * **Distributed workflows** - Delegate tasks to specialized agent instances * **Context isolation** - Maintain separate workspaces for different agents * **Long-running collaboration** - Persistent agent-to-agent conversations * **Project-specific delegation** - Route work to agents with specific project contexts ### Step 1: Enable A2A Server 1. Click **Settings** in the sidebar 2. Navigate to the **MCP/A2A** tab 3. Scroll to the **A0 A2A Server** section 4. Toggle **Enable A2A server** to ON ![A2A Server Settings](https://www.agent-zero.ai/res/docs/setup/a2a/a2a-conn.png) ### Step 2: Get Connection URL 1. Click on **connection example** to view your A2A connection details 2. The dialog displays: * **API Token** - Automatically generated from your username and password * **A2A Connection URL** - The full URL other agents will use to connect * Optional **Project selector** - To create project-specific connection URLs ![A2A Connection Dialog](https://www.agent-zero.ai/res/docs/setup/a2a/a2a2.png) ### Step 3: Save Configuration 1. Click **Save** to apply your settings 2. The A2A server is now active and ready to accept connections **⚠️ Important:** The API token changes when you update your Agent Zero credentials. Existing connections will need to be reconfigured with the new token. ### Connection URL Format #### Basic A2A Connection http://YOUR_HOST:PORT/a2a/t-YOUR_API_TOKEN #### With Project Context To connect with a specific project active: http://YOUR_HOST:PORT/a2a/t-YOUR_API_TOKEN/p-PROJECT_NAME When a project is specified, all A2A conversations run in that project's context with access to project-specific resources and knowledge. ### Example Use Cases #### 1\. Local Development Setup Two Agent Zero instances on the same machine: Instance 1: http://localhost:8080/a2a/t-abc123xyz Instance 2: http://localhost:8081/a2a/t-def456uvw #### 2\. Remote Agent Collaboration Connect to a remote Agent Zero instance: http://agent.example.com:8080/a2a/t-remote-token #### 3\. Project-Specific Delegation Main agent delegates frontend work to specialized agent: http://localhost:8081/a2a/t-frontend-token/p-webapp-ui ### A2A vs MCP Comparison | Feature | A2A | MCP | | --- | --- | --- | | **Purpose** | Agent-to-agent chat delegation | Tool/function access | | **Use Case** | Long-running conversations | Specific tool calls | | **Context** | Full chat context | Function parameters only | | **Best For** | Workflow delegation | Tool integration | **💡 Tip:** Use A2A when you need another agent's reasoning and conversation capabilities. Use MCP when you just need access to specific tools or functions. ### Security Considerations * **Token Protection:** Keep your API tokens secure - they provide full access to your Agent Zero instance * **Network Access:** Consider using firewalls or reverse proxies to restrict A2A endpoint access * **HTTPS:** For production deployments, use HTTPS to encrypt A2A communication * **Credential Rotation:** Changing your password will invalidate all existing A2A connection URLs ### Testing Your Connection You can test A2A connectivity using curl: curl -X POST http://localhost:8080/a2a/t-YOUR_TOKEN \ -H "Content-Type: application/json" \ -d '{"message": "Hello from another agent"}' Troubleshooting --------------- ### MCP Issues #### Server Not Connecting * Verify the A2A server is enabled in Settings * Check that the Agent Zero instance is running * Confirm the port is accessible (check firewall rules) * Wrong URL or port configuration * Missing authentication headers for remote servers #### Tools Not Appearing * Confirm server shows as connected (green status) * Check server exposes tools (count shown in UI) * Verify tool names match server documentation * For `npx`/`uvx` servers, first run downloads packages (may take time) ### A2A Issues #### Connection Refused * Verify the A2A server is enabled in Settings * Check that the Agent Zero instance is running * Confirm the port is accessible #### Invalid Token * Token may have changed due to credential updates * Generate a new connection URL from Settings → MCP/A2A * Update the connecting agent's configuration #### Project Not Found * Verify the project name in the URL matches exactly * Check that the project exists in the target instance * Project names are case-sensitive Need Help? ---------- Questions about MCP or A2A integration? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Multi-Agent Cooperation - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Subagents Architecture ============ Multi-Agent Cooperation ----------------------- Agent Zero employs hierarchical multi-agent cooperation where superior agents delegate tasks to subordinates for efficient problem-solving and workload distribution. Understanding Multi-Agent Architecture -------------------------------------- One of Agent Zero's unique features is multi-agent cooperation. The framework uses a hierarchical structure where agents can create and delegate to subordinate agents, enabling complex task decomposition and parallel execution. ![Multi-Agent Cooperation](https://www.agent-zero.ai/res/docs/usage/multi-agent.png) ### Key Concepts * **Creating Sub-Agents:** Agents can create subordinate agents to delegate sub-tasks, helping manage complexity and distribute workload * **Communication:** Agents can communicate with each other, sharing information and coordinating actions through the message system * **Hierarchy:** Agent Zero uses a hierarchical structure with superior agents delegating tasks to subordinates for structured problem-solving Agent Hierarchy and Communication --------------------------------- The user or Agent 0 is at the top of the hierarchy, delegating tasks to subordinate agents, which can further delegate to other agents. Each agent can utilize tools and access shared assets (prompts, memory, knowledge, extensions, and skills) to perform its tasks. ### Communication Flow Communication flows between agents through messages structured according to prompt templates. Messages typically include: * **Thoughts:** The agent's Chain of Thought and planning process * **Tool name:** The specific tool used by the agent * **Responses or queries:** Results, feedback, or queries from tools or other agents ### Typical Interaction Flow 1. The user provides an instruction to Agent 0 2. Agent 0 initializes VectorDB and accesses memory 3. Agent 0 analyzes the instruction and formulates a plan using `thoughts`, possibly involving tools or sub-agents 4. If necessary, Agent 0 delegates sub-tasks to subordinate agents 5. Subordinate agents use tools to perform actions, providing arguments and responses 6. Agents communicate results and feedback back up the hierarchy 7. Agent 0 provides the final response to the user **💡 Key Benefit:** This hierarchical approach allows efficient breakdown of complex tasks into smaller, manageable sub-tasks with parallel execution where possible. The call\_subordinate Tool -------------------------- Agents use the `call_subordinate` tool to delegate tasks to subordinate agents. This is Agent Zero's primary mechanism for task decomposition. ### How It Works 1. Agent 0 identifies a sub-task that can be delegated 2. Uses `call_subordinate` tool with specific instructions 3. A new subordinate agent is spawned with its own context 4. Subordinate agent executes the task using available tools 5. Results are communicated back to the parent agent 6. Parent agent integrates results into its overall solution **Example Delegation:** Agent 0: "I need to analyze a large dataset and generate visualizations" ↓ delegates to → Subordinate 1: "Analyze the dataset and extract key metrics" Subordinate 2: "Generate charts and visualizations from the metrics" ↓ results flow back to → Agent 0: Combines results into final report Agent Profiles & Configuration ------------------------------ Agent profiles allow you to customize the behavior of Agent 0 and its subordinates by overriding prompts, tools, extensions, and sub-agent settings. Each profile can define unique characteristics and capabilities. ### What Are Agent Profiles? Agent profiles are custom configurations located in `/agents//` that modify how agents behave. They can override: * **Prompts:** Customize system prompts and communication style * **Tools:** Add custom tools or override existing ones * **Extensions:** Modify agent lifecycle hooks and behavior * **Sub-agent settings:** Control which sub-agents are available (via `agents.json`) ### Profile Structure /agents/ └── my_custom_agent/ ├── prompts/ # Override system prompts │ ├── agent.system.main.role.md │ ├── agent.system.main.communication.md │ └── agent.system.tool.*.md ├── tools/ # Custom or overridden tools │ └── custom_tool.py ├── extensions/ # Custom extensions │ └── agent_init/ │ └── custom_extension.py └── agents.json # Sub-agent configuration ### agents.json Configuration The `agents.json` file controls which sub-agents are enabled or disabled for a specific agent profile. This allows you to customize the available tools and capabilities for subordinate agents. **Example agents.json:** { "browser_agent": { "enabled": true }, "code_agent": { "enabled": true }, "research_agent": { "enabled": false } } When a subordinate agent is created, it inherits the sub-agent configuration from the parent agent's profile, ensuring consistent behavior across the agent hierarchy. ### Creating a Custom Agent Profile 1. Create a new directory under `/agents/` with your profile name 2. Add only the components you want to customize (prompts, tools, extensions) 3. Create `agents.json` to configure available sub-agents 4. Select your profile in Settings → Agent Configuration → Agent Profile **💡 Tip:** Start by cloning the `_example` agent profile and modifying it to your needs. Agent Zero automatically merges your custom profile with the default configuration. ### Built-in Profiles Agent Zero includes several pre-configured profiles: * **agent0:** Default profile with balanced capabilities * **hacker:** Security-focused profile with specialized prompts and tools * **researcher:** Optimized for research and information gathering * **\_example:** Template profile for creating custom configurations ### Project-Specific Agent Configuration When using [Projects](https://www.agent-zero.ai/p/docs/projects/) , each project can have its own `agents.json` file located in `.a0proj/agents.json`. This allows project-specific sub-agent configurations: * Enable/disable specific sub-agents per project * Different capabilities for different project types * Isolated sub-agent behavior within project context Project-level `agents.json` takes precedence over the agent profile's configuration when a project is active. Common Multi-Agent Use Cases ---------------------------- ### Complex Task Decomposition Break large tasks into specialized sub-tasks: * Research agent gathers information * Analysis agent processes data * Writing agent generates reports * Review agent validates output ### Parallel Processing Execute independent tasks simultaneously: * Multiple data sources processed in parallel * Independent API calls delegated to separate agents * Concurrent file processing operations * Results aggregated by parent agent ### Specialized Agent Roles Delegate to agents with specific expertise: * Coding agent for implementation tasks * Testing agent for quality assurance * Documentation agent for writing guides * DevOps agent for deployment operations Best Practices -------------- ### When to Use Sub-Agents * **Complex tasks:** When a task can be broken into clear sub-components * **Parallel work:** When independent operations can run simultaneously * **Context separation:** When sub-tasks benefit from isolated context * **Specialization:** When different expertise is needed for different parts ### Avoid Over-Delegation * Don't delegate trivial tasks - adds overhead * Keep delegation depth reasonable (2-3 levels max) * Ensure sub-tasks are sufficiently independent * Consider token costs of multiple agent contexts ### Communication Guidelines * Provide clear, specific instructions to subordinates * Define expected output format * Include necessary context but avoid redundancy * Use structured message formats for consistency Need Help? ---------- Questions about multi-agent cooperation? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Secrets & Variables - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Secrets Security ======== Secrets & Variables ------------------- Manage secrets and environment variables securely for API keys, credentials, and configuration values. Agent Zero supports both **global secrets** (Settings) and **project-scoped secrets** (Projects). For detailed information on project-specific secrets and variables, see the Projects guide. [→ View Project Secrets Guide](https://www.agent-zero.ai/p/docs/projects/#configuration) ### Global Secrets Global secrets are available to all chats and projects unless overridden: 1. Open **Settings** in the sidebar 2. Navigate to **External Services** or **Agent Settings** 3. Enter API keys for services (OpenRouter, OpenAI, etc.) 4. Click **Save** **Common Global Secrets:** * OpenRouter API Key * OpenAI API Key * Anthropic API Key * Database credentials * Email service credentials ### Project-Scoped Secrets Projects can have their own isolated secrets that override global settings: #### Variables (Non-sensitive) Stored in `.a0proj/variables.env`: API_BASE_URL=https://api.example.com DEFAULT_TIMEOUT=30 OUTPUT_FORMAT=json #### Secrets (Sensitive) Stored in `.a0proj/secrets.env`: API_KEY=sk-abc123xyz... DATABASE_PASSWORD=super_secret_pwd GITHUB_TOKEN=ghp_xyz789... **⚠️ Security:** Project secrets override global secrets and are stored locally in the container. They are masked in the UI and may not be included in automatic backups. ### Using Secrets in Prompts Reference secrets by name in your prompts: **Example:** "Use my API_KEY to authenticate with the external service and fetch the data." "Connect to the database using my DATABASE_PASSWORD and run the query." The agent automatically accesses the appropriate secrets from the current context (project-scoped if available, otherwise global). ### Best Practices * **Use project secrets** for client-specific credentials * **Never commit secrets** to version control * **Backup secrets manually** - they may not be in automatic backups * **Rotate credentials** periodically for security * **Use meaningful names** that clearly indicate purpose * **Document required secrets** in project instructions We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Community - Agent Zero [Home](https://www.agent-zero.ai/) › [Docs](https://www.agent-zero.ai/p/docs/) › Community Connect ======= Join the Community ------------------ Connect with other Agent Zero users, share knowledge, get support, and contribute to the growing ecosystem. Getting Help ------------ ### Before Asking 1. Check the [documentation](https://www.agent-zero.ai/p/docs/) first 2. Search existing Discord threads and Skool posts 3. Review GitHub issues for similar problems 4. Have your configuration and error logs ready ### When Asking for Help * **Be specific:** Describe what you're trying to achieve * **Include context:** Share relevant configuration and code * **Show errors:** Provide complete error messages and logs * **Explain attempts:** What have you tried so far? * **Be patient:** Community members volunteer their time Contributing ------------ Agent Zero thrives on community contributions. Whether you're a developer, writer, or enthusiast, there are many ways to contribute: * **Code contributions:** Submit bug fixes, features, or tools * **Documentation:** Improve guides, tutorials, and examples * **Community support:** Help other users on Discord and Skool * **Share projects:** Showcase what you've built with Agent Zero * **Feedback:** Report issues and suggest improvements Connect Now ----------- **Ready to join the Agent Zero community?** * [Discord Server](https://discord.gg/B8KZKNsPpj) - Real-time chat and support * [Skool Community](https://www.skool.com/agent-zero) - Structured learning and discussions * [GitHub](https://github.com/agent0ai/agent-zero) - Code, issues, and contributions We use essential cookies for login and saving your preferences. With your consent, we also use analytics cookies (Google Analytics) to improve the website. [Learn more](https://www.agent-zero.ai/p/cookies/) [Individual options](https://www.agent-zero.ai/p/cookies/) Accept all cookies --- # Working with Projects - Agent Zero User Guides =========== Working with Projects --------------------- Create isolated workspaces with dedicated context, instructions, memory, and secrets for organized multi-client or multi-domain workflows. What Are Projects? ------------------ Projects are one of Agent Zero's most powerful features for organizing your work. They create isolated workspaces with dedicated context, instructions, memory, and secrets - preventing context bleed between unrelated tasks or clients. ### A project in Agent Zero includes: * **Dedicated folder structure** under `/a0/usr/projects//` * **Custom instructions** automatically injected into the agent's system prompt * **Isolated or shared memory** to prevent context pollution * **Project-specific secrets and variables** for secure credential management * **Custom agent configurations** including subagent settings * **Knowledge base integration** with project-scoped documents * **Git repository integration** for version-controlled codebases * **File structure injection** for automatic codebase awareness **💡 Think of it:** Projects are separate "personas" or "modes" for your agent - each with its own context, memory, and capabilities. Creating Your First Project --------------------------- ### Access Methods #### Method 1: Dashboard Access 1. Open the **Dashboard** from the sidebar 2. Click the **Projects** button 3. The Projects modal will open #### Method 2: Quick Access 1. Look at the top-right corner where it shows **"No project"** 2. Click on it to open the project dropdown 3. Click **"Projects"** to open the modal ![Project Activation Dropdown](https://www.agent-zero.ai/res/docs/usage/projects/projects-activate-project.png) ### Creating an Empty Project Once in the Projects modal: 1. Click **"Create project"** button 2. You'll see the project creation wizard ![Project Creation Form](https://www.agent-zero.ai/res/docs/usage/projects/projects-creation.png) #### The wizard includes: * **Title:** A human-readable name displayed in the UI * **Color tag:** Visual identifier for quick recognition * **Git Repository (optional):** URL to clone a repository **💡 Tip:** Choose meaningful titles and distinct colors, especially if you plan to work with multiple projects. This helps you quickly identify which project is active. ### Creating a Git-Based Project Agent Zero's **Git Projects** feature allows you to clone repositories directly into your project workspace: 1. In the creation form, enter a Git repository URL in the **Git Repository** field * Supports both public and private repositories * For private repositories, you can provide an authentication token 2. Click **"Create and continue"** 3. Agent Zero will clone the repository ![Git Clone Progress](https://www.agent-zero.ai/res/docs/usage/projects/projects-gitprojects-clone.png) #### The cloning process: * Downloads the repository to `/a0/usr/projects//` * By default, clones the `main` branch * Supports authentication via token (stored securely, never in URLs) * You can ask the agent to checkout different branches later ![Git Project Status](https://www.agent-zero.ai/res/docs/usage/projects/projects-git-projects-tree.png) After cloning completes, you'll see repository URL, current branch, clean/dirty status, and last commit information. **📝 Note:** If the cloned repository already contains a `.a0proj/` folder with `project.json`, Agent Zero will merge the existing configuration with your specified title and color preferences. Project Configuration --------------------- Once created, you can configure your project through the Edit Project interface: ![Project Description and Instructions](https://www.agent-zero.ai/res/docs/usage/projects/projects-desc-and-instructions.png) ### Description and Instructions #### Description The **Description** field helps both you and the agent understand the project's purpose. This is purely informational and helps provide context when switching between projects. **Example:** An intelligent Excel automation system that processes financial spreadsheets from multiple sources, performs data validation and cleaning, consolidates information, generates executive reports, and flags anomalies or compliance issues. #### Instructions The **Instructions** field is critical - these are automatically injected into the agent's system prompt when the project is active. Write clear, specific instructions that guide the agent's behavior for this project. **Example Instructions:** ## Your Role You are an expert Financial Data Analyst and Excel Automation Specialist. Your job is to process, analyze, and report on financial spreadsheet data with precision and business acumen. ## Operational Context - Work directory: `/usr/projects/excel-finance/` - Input data location: `/usr/projects/excel-finance/data/incoming/` - Processed files: `/usr/projects/excel-finance/data/processed/` - Reports output: `/usr/projects/excel-finance/reports/` ## Core Responsibilities ### 1. Data Ingestion & Discovery - Scan `/data/incoming/` for new Excel/CSV files - Identify data types, sheets, and file structures - Log source details and timestamps ### 2. Data Validation & Cleaning - Check for missing values, duplicates, format inconsistencies - Validate numerical ranges and date formats - Flag suspicious patterns or outliers - Document all cleaning operations ### 3. Executive Reporting - Generate summary reports using templates - Include key metrics, trends, and visualizations - Highlight anomalies and compliance issues **⚠️ Important:** Well-written instructions are the key to effective project-specific agent behavior. Be specific about: * The agent's role and expertise level * Directory paths and file locations * Expected workflows and procedures * Quality standards and validation rules * Output formats and requirements ### Memory Isolation Projects support two memory modes: * **Own memory** (default): The project has its own isolated memory directory * **Global memory**: The project shares the default memory pool **💡 Use own memory for:** * Client projects with confidential information * Different programming languages/frameworks * Distinct domains (e.g., finance vs. marketing) **Use global memory for:** * Related sub-projects in the same domain * When you want the agent to remember across all work * Personal projects without sensitive data ### Variables and Secrets Projects support scoped configuration values: #### Variables (Non-sensitive) Stored in `.a0proj/variables.env`: API_BASE_URL=https://api.example.com DEFAULT_TIMEOUT=30 OUTPUT_FORMAT=json LOG_LEVEL=info #### Secrets (Sensitive) Stored in `.a0proj/secrets.env`: API_KEY=sk-abc123xyz... DATABASE_PASSWORD=super_secret_pwd GITHUB_TOKEN=ghp_xyz789... SMTP_PASSWORD=email_pwd_here **🔒 Security Notes:** * Secrets are stored locally in the container * They are masked in the UI when editing * Reference secrets in prompts by name: "Use my `API_KEY` to authenticate" Activating a Project -------------------- To use a project, you need to activate it for your current chat: ### In a New Chat 1. Activate a project from the Projects modal or dropdown 2. Start a new chat 3. The project is automatically active for that chat ### In an Existing Chat 1. Click the project dropdown in the top-right corner (where it shows current project or "No project") 2. Select the desired project from the list 3. The project activates immediately for this chat ![Project Activation](https://www.agent-zero.ai/res/docs/usage/projects/projects-activate-project.png) **📝 Note:** Each chat can have its own active project. You can have multiple chats open with different projects active simultaneously. ### First Interaction After Activation Once activated, the agent's behavior changes according to the project's configuration: ![First Operations in Project](https://www.agent-zero.ai/res/docs/usage/projects/projects-first-ops.png) #### The agent now: * Has access to project-specific instructions * Uses project-isolated memory (if configured) * Can reference project secrets and variables * Sees the project file structure (if enabled) * Operates within the project's working directory **Example prompt after activation:** Hi A0. Start first development phase: - recon - scope definition - AGENTS.md generation The agent will understand the context based on the project's instructions and work accordingly. Common Use Cases ---------------- ### Multi-Client Agency Work Use projects to isolate each client: /usr/projects/client-acme/ /usr/projects/client-globex/ /usr/projects/client-initech/ Each with client-specific instructions, isolated memory, API keys, and knowledge bases. ### Multi-Language Development Create projects for different tech stacks: /usr/projects/python-ml-research/ /usr/projects/nodejs-webapp/ /usr/projects/go-microservices/ Each with language-specific code patterns, framework documentation, and linting rules. ### Domain-Specific Work Organize by business domain: /usr/projects/finance-automation/ /usr/projects/marketing-analytics/ /usr/projects/hr-automation/ Each with domain-specific terminology, industry standards, and specialized tools. Best Practices -------------- ### Instruction Writing * **Be specific:** Don't say "analyze data" - specify format, methods, and output * **Include paths:** Always mention where files should be read/written * **Define quality standards:** What constitutes "good" output? * **Specify error handling:** How should the agent respond to issues? * **Use markdown formatting:** Clear sections and lists improve readability ### Memory Management * Use **own memory** by default for client work * Use **global memory** only when projects truly need shared knowledge * Regularly review and clean project memory via the Memory Dashboard * Clear memory when archiving completed projects ### Security * Never commit `.a0proj/secrets.env` to version control * Use project-scoped secrets instead of global secrets for sensitive work * Consider separate projects for different security levels * Backup secrets manually - they may not be included in automatic backups ### Organization * Use clear, descriptive project names * Choose distinct colors for visual identification * Keep instructions updated as project scope evolves * Document any manual setup steps in project instructions Need Help? ---------- Questions about projects? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Updating Agent Zero - Agent Zero Maintenance =========== Updating Agent Zero ------------------- Keep your Agent Zero installation up to date with the latest features and improvements. For comprehensive update instructions including safe update procedures and backup strategies, see the **Updating Agent Zero** section in the Get Started guide. [→ View Update Guide](https://www.agent-zero.ai/p/docs/get-started/#updating) ### Quick Update Steps 1. **Keep current container running** - Don't stop it yet! 2. **Pull latest image:** `docker pull agent0ai/agent-zero:latest` 3. **Run NEW container on different port:** `docker run -d -p 50081:80 --name agent-zero-new agent0ai/agent-zero` 4. **Backup from OLD instance:** Settings → Backup & Restore → Create Backup 5. **Restore to NEW instance:** Settings → Backup & Restore → Restore 6. **Verify and cleanup:** Only delete old container after confirming data is restored **💡 Why This Process?** Running both containers ensures you can always access your current data. If something goes wrong with the new version, your original Agent Zero is still running with all your data intact. ### Additional Resources * [Data Persistence & Backups](https://www.agent-zero.ai/p/docs/get-started/#data-persistence) * [GitHub Releases](https://github.com/agent0ai/agent-zero/releases) * [Changelog](https://www.agent-zero.ai/p/docs/changelog/) --- # Memory Management - Agent Zero System Features =============== Memory Management ----------------- Agent Zero includes a sophisticated memory system that stores and retrieves information from conversations, knowledge sources, and learning experiences through its Memory Dashboard. What is Agent Memory? --------------------- Agent Zero's memory system uses vector embeddings to store, organize, and retrieve information. This enables the agent to remember conversations, learn from interactions, and access imported knowledge efficiently. ### Memory Types * **Conversation Memory:** Stores chat history and interactions * **Knowledge Base:** Imported documents and reference materials * **Learning Memory:** Patterns and insights gathered from tasks * **Project Memory:** Project-specific isolated or shared memory **💡 Key Feature:** The Memory Dashboard provides a powerful interface to view, search, filter, edit, and delete memory entries stored in the vector database. Accessing the Memory Dashboard ------------------------------ 1. Click the **Memory** button in the sidebar 2. The dashboard displays all memory entries with their metadata and preview 3. Memory entries are organized by directory (e.g., "default") and area (e.g., "All Areas", "main") ![Memory Dashboard](https://www.agent-zero.ai/res/docs/usage/memory-dashboard.png) Dashboard Features ------------------ ### Filtering and Organization The dashboard provides several ways to organize and filter memory entries: * **Memory Directory:** Select which memory directory to view (default or project-specific) * **Area Filter:** Filter memories by area to focus on specific contexts * **Limit Control:** Adjust the number of results displayed (default: 1000) #### Statistics Bar View counts at a glance: * **Total:** All memory entries in the directory * **Filtered:** Entries matching current filters * **Knowledge:** Entries from knowledge sources * **Conversation:** Entries from chat interactions ### Search Functionality Find specific memories using the powerful search feature: * **Search Bar:** Enter keywords to search memory content * **Threshold Slider:** Adjust search sensitivity (0.00 - 1.00) * Lower values return more results with looser matching * Higher values return fewer, more precise matches * Default threshold: 0.60 * **Real-time Search:** Results update as you type **💡 Tip:** Use the threshold slider to balance between broad discovery (lower values) and precise matching (higher values). Start with 0.60 and adjust based on your results. ### Memory Entry Information Each memory entry displays key information: * **Tags:** Visual indicators showing the memory type (MAIN, Knowledge, etc.) * **Timestamp:** When the memory was created (format: MM/DD HH:MM) * **Preview:** First few lines of the memory content * **Action Icons:** Quick access to view details, edit, or delete Viewing and Editing Memories ---------------------------- Click any memory entry to open the detailed view with full editing capabilities: ![Memory Details](https://www.agent-zero.ai/res/docs/usage/memory-editing.png) ### Memory Details View The details panel provides comprehensive information: #### Memory Content * Full text of the stored memory * Syntax highlighting for structured content * Editable text area for modifications #### Metadata Panel * Unique ID * Area classification * Source type (e.g., "Knowledge") * Timestamp (creation date and time) * Source file (if applicable) ### Editing Memory Content 1. Click on a memory entry to open the details view 2. Edit the content in the text area 3. Click **"Save Changes"** to update the memory 4. The memory is re-embedded with updated content **⚠️ Note:** Editing memory content will regenerate its vector embeddings. This may affect how the agent retrieves this memory in future interactions. Memory Management Operations ---------------------------- ### Deleting Memories 1. Select memory entries you want to delete (checkbox next to each entry) 2. Click the **"Delete Selected"** button 3. Confirm the deletion 4. Selected memories are permanently removed from the vector database **⚠️ Warning:** Deleted memories cannot be recovered. Use caution when deleting, especially for conversation history or imported knowledge. ### Bulk Operations * **Select All:** Checkbox in header selects all visible entries * **Filter + Delete:** Apply filters, select all, then delete to clean specific memory types * **Export:** Future feature for backing up memories ### Memory Cleanup Strategies #### When to clean memory: * Agent performance degrades due to too many irrelevant memories * Project completion - archive or remove project-specific memories * Context pollution - agent retrieves outdated or incorrect information * Privacy concerns - remove sensitive conversation history #### Best practices: * Use project-specific memory to isolate client or domain data * Periodically review and clean old conversation memories * Keep knowledge base memories organized by source * Use meaningful area classifications for easier filtering Project Memory Integration -------------------------- ### Memory Isolation with Projects When a project has its own memory enabled, the Memory Dashboard shows project-specific directories: * Select project memory directory from dropdown * View only memories associated with that project * Prevents context bleed between clients or domains * Manage project knowledge separately **💡 Tip:** For more information about project memory isolation and configuration, see the [Projects Guide](https://www.agent-zero.ai/p/docs/projects/) . Knowledge Base Management ------------------------- ### Importing Knowledge 1. Click the **"Import Knowledge"** button in the action bar 2. Select files to import (.txt, .pdf, .csv, .html, .json, .md) 3. Files are processed and stored in `/a0/knowledge/custom/main` 4. Content is automatically embedded and added to the vector database **💡 Tip:** Imported knowledge is tagged as "Knowledge" type in the Memory Dashboard, making it easy to filter and manage separately from conversation memories. ### Managing Knowledge Entries * **Filter by type:** Use area filter to show only "Knowledge" memories * **Edit content:** Update imported knowledge if source material changes * **Remove outdated:** Delete obsolete knowledge to prevent confusion * **Organize by source:** Use naming conventions in source files for easier management Best Practices -------------- ### Memory Organization * Use project memory isolation for client work * Regularly review and clean conversation history * Keep knowledge base updated and relevant * Use meaningful area classifications * Balance memory size with performance ### Search Optimization * Start with threshold 0.60, adjust based on results * Use specific keywords for better retrieval * Filter by area before searching for faster results * Verify search results match your intent ### Maintenance * Schedule periodic memory reviews * Remove completed project memories after archiving * Clear test or debugging memories regularly * Monitor memory database size * Use backup/restore before major cleanups Need Help? ---------- Questions about memory management? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Skills - Agent Zero Development =========== Skills ------ Create contextual expertise modules using the SKILL.md standard. Skills provide token-efficient, cross-platform capabilities that are loaded only when relevant. What Are Skills? ---------------- A **Skill** is a contextual expertise module that provides the AI agent with specialized knowledge and procedures for specific tasks. Unlike tools (which are always loaded), skills are **surfaced via description/tag matching** when relevant, making them token-efficient and context-aware. ### Skills vs Tools vs Knowledge | Aspect | Skills | Tools | Knowledge | | --- | --- | --- | --- | | **Loading** | Description/tag matching | Always in prompt | Semantic recall | | **Purpose** | Procedures & expertise | Actions & functions | Facts & data | | **Format** | SKILL.md (YAML + Markdown) | Python/code | Text/documents | | **When to use** | "How to do X" | "Do X now" | "What is X" | **💡 Key Benefit:** Skills are indexed in vector memory and loaded dynamically when needed, keeping your system prompt clean and token-efficient while providing contextual expertise. Cross-Platform Compatibility ---------------------------- The SKILL.md standard (originally developed by Anthropic) is compatible with multiple AI platforms. Skills you create can be used across: * **Agent Zero** (this project) * **Claude Code** (Anthropic) * **Cursor** (AI IDE) * **OpenAI Codex CLI** * **GitHub Copilot** * **Goose** (Block) Skills from other platforms can be copied directly to Agent Zero's `usr/skills/` directory and will work without modification. SKILL.md Standard ----------------- Every skill must have a `SKILL.md` file with YAML frontmatter followed by markdown content: ### Basic Structure --- name: "skill-name" description: "Clear, concise description of what this skill does and when to use it" version: "1.0.0" author: "Your Name " license: "MIT" tags: ["category", "purpose", "technology"] triggers: - "keyword that activates this skill" - "another trigger phrase" allowed_tools: - tool_name - another_tool metadata: complexity: "beginner|intermediate|advanced" category: "development|devops|data|productivity|creative" --- # Skill Name ## Overview Brief description of what this skill accomplishes. ## When to Use - Situation 1 where this skill applies - Situation 2 where this skill applies ## Instructions ### Step 1: First Step Detailed instructions... ### Step 2: Second Step More instructions... ## Examples ### Example 1: Basic Usage \`\`\`python # Code example \`\`\` ## Common Pitfalls - Pitfall 1 and how to avoid it ## Related Skills - [related-skill-1](../related-skill-1/SKILL.md) ### Required Fields | Field | Description | | --- | --- | | `name` | Unique identifier (lowercase, hyphens allowed) | | `description` | What the skill does (used for semantic matching) | ### Optional Fields | Field | Description | | --- | --- | | `version` | Semantic version (e.g., "1.0.0") | | `author` | Your name and email | | `license` | License (MIT, Apache-2.0, etc.) | | `tags` | Categories for discovery | | `triggers` | Phrases that activate this skill | | `allowed_tools` | Tools this skill can use | | `metadata` | Additional structured data | Creating Your First Skill ------------------------- ### Using the CLI (Recommended) # Create a new skill interactively python -m python.helpers.skills_cli create my-skill-name # List all available skills python -m python.helpers.skills_cli list # Validate a skill python -m python.helpers.skills_cli validate my-skill-name # Search skills python -m python.helpers.skills_cli search "keyword" ### Manual Creation 1. Create a folder in `usr/skills/` with your skill name 2. Add a `SKILL.md` file with YAML frontmatter 3. Optionally add supporting scripts (`.py`, `.sh`, `.js`) my-awesome-skill/ ├── SKILL.md # Required ├── helper.py # Optional Python script ├── setup.sh # Optional shell script └── templates/ # Optional templates folder └── config.json ### Directory Structure | Directory | Description | | --- | --- | | `/skills` | Default skills included with Agent Zero | | `/usr/skills` | Your custom skills (create folders here) | Using Skills ------------ Skills are automatically discovered and loaded when contextually relevant. The agent uses the `skills_tool` to: * **List all available skills** - Browse the skills catalog * **Load a specific skill by name** - Explicitly activate a skill * **Read files from within a skill directory** - Access helper scripts and templates **💡 Automatic Discovery:** When you ask the agent to perform a task, it searches for relevant skills based on description and tag matching. You don't need to explicitly load them. Best Practices -------------- ### Writing Effective Descriptions The `description` field is crucial for semantic matching. Make it specific and include synonyms: **Good:** description: "Guides systematic debugging of Python applications using print statements, debugger, and logging to identify root causes" **Bad:** description: "Helps with debugging" ### Structuring Instructions * **Be Specific** - Avoid vague instructions * **Use Steps** - Number your steps clearly * **Include Examples** - Show, don't just tell * **Anticipate Errors** - Include troubleshooting ### Keep Skills Focused One skill = one expertise area. If your skill is getting too long, split it: * `api-design` - API structure and patterns * `api-security` - API authentication and authorization * `api-testing` - API testing strategies ### Naming Conventions * Use lowercase with hyphens: `my-skill-name` * Be descriptive: `python-debugging` not `debug` * Avoid generic names: `fastapi-crud` not `api` Sharing Skills -------------- ### Contributing to Agent Zero 1. Fork the [Agent Zero repository](https://github.com/agent0ai/agent-zero) 2. Create your skill in `usr/skills/` 3. Move to `skills/` for contribution: `mv usr/skills/my-skill skills/my-skill` 4. Create a Pull Request with branch name: `feat/skill-my-skill-name` ### Publishing to Skills Marketplace Share your skills on [skillsmp.com](https://skillsmp.com/) : 1. Create a GitHub repository for your skill 2. Ensure it follows the SKILL.md standard 3. Submit to the marketplace via their contribution process Frequently Asked Questions -------------------------- ### Where should I put my skills? During development, use `usr/skills/`. For contribution to Agent Zero, move to `skills/`. ### How are skills discovered? Skills are matched against their name, description, and tags for the current query. They are indexed in vector memory for semantic search. ### Can I use skills from other platforms? Yes! The SKILL.md standard is cross-platform. Skills from Claude Code, Cursor, or other compatible platforms can be copied directly to `usr/skills/`. ### How do I update a skill? Edit the SKILL.md file and increment the version number. Changes take effect on agent restart. ### Can skills call other skills? Skills don't directly call each other, but the agent may combine multiple skills when appropriate for a task. Need Help? ---------- Questions about creating or using skills? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Get Started - Agent Zero Documentation ============= Get Started with Agent Zero --------------------------- Install, configure, and start using Agent Zero in minutes. Watch the walkthrough or follow the steps below. [![Agent Zero - Easy Installation Guide](https://www.agent-zero.ai/res/sd/w5v5Kjx51hs-sddefault.webp) ![Play video](https://www.agent-zero.ai/res/YT-play.svg)](https://www.youtube.com/watch?v=w5v5Kjx51hs) How to Install Agent Zero ------------------------- Agent Zero runs in a secure, isolated Docker container, which is the **recommended and safest method** for deployment. This makes installation simple and consistent across all platforms — Windows, macOS, and Linux. ### Step 1: Install Docker Desktop Docker Desktop is free for personal use and required to run Agent Zero. ![Windows](https://www.agent-zero.ai/res/windows.png) Windows![macOS](https://www.agent-zero.ai/res/apple.png) macOS ![Linux](https://www.agent-zero.ai/res/linux.png) Linux 1. **Download Docker Desktop:** Visit [docker.com](https://www.docker.com/products/docker-desktop) and download Docker Desktop for Windows ![Docker Desktop download page](https://www.agent-zero.ai/res/docs/setup/image-8.png) 2. **Run the installer:** Double-click the downloaded file and follow the installation wizard with default settings ![Docker installer](https://www.agent-zero.ai/res/docs/setup/image-9.png) 3. **Launch Docker Desktop:** After installation, start Docker Desktop from your Start menu ![Docker Desktop Windows icon](https://www.agent-zero.ai/res/docs/setup/image-11.png) 4. **Verify installation:** Wait for Docker Desktop to fully start (you'll see the Docker icon in your system tray) **💡 Note:** Docker Desktop on Windows requires WSL 2. The installer will set this up automatically if needed. 1. **Download Docker Desktop:** Visit [docker.com](https://www.docker.com/products/docker-desktop) and download Docker Desktop for Mac (choose your chip: Intel or Apple Silicon) 2. **Install the app:** Open the downloaded .dmg file and drag Docker to your Applications folder ![Docker Desktop macOS installation](https://www.agent-zero.ai/res/docs/setup/image-13.png) 3. **Launch Docker Desktop:** Open Docker from your Applications folder 4. **Configure socket (important!):** On first launch, go to Settings → Advanced and ensure "Allow the default Docker socket to be used" is checked ![Docker socket configuration](https://www.agent-zero.ai/res/docs/setup/macsocket.png) **💡 Note:** Make sure to allow Docker file system access when prompted by macOS. 1. **Install Docker CE:** Use your package manager or follow the [official Docker guide](https://docs.docker.com/desktop/install/linux-install/) # Ubuntu/Debian sudo apt-get update sudo apt-get install docker-ce docker-ce-cli containerd.io # Fedora sudo dnf install docker-ce docker-ce-cli containerd.io # Arch sudo pacman -S docker 2. **Add user to docker group:** This allows running Docker without sudo sudo usermod -aG docker $USER newgrp docker # Or log out and back in 3. **Start Docker service:** sudo systemctl start docker sudo systemctl enable docker # Start on boot 4. **Verify installation:** docker --version docker ps # Should run without errors ### Step 2: Pull the Agent Zero Image The Agent Zero Docker image contains a complete Linux system with all necessary software pre-installed. It's approximately 2GB to download and 4.5GB when extracted. #### Option A: Using Docker Desktop (Recommended) 1. Open Docker Desktop 2. Click on the search bar at the top 3. Search for `agent0ai/agent-zero` 4. Select the image (should show 10K+ downloads) 5. Click "Pull" to download the latest version ![Docker image search](https://www.agent-zero.ai/res/docs/setup/1-docker-image-search.png) #### Option B: Using Command Line docker pull agent0ai/agent-zero:latest _The download will take a few minutes depending on your internet connection._ ### Step 3: Run Your Agent Zero Container Docker images are blueprints; containers are running instances. You can run multiple independent Agent Zero containers from a single image if needed. #### Option A: Using Docker Desktop (Recommended) 1. Go to the "Images" tab in Docker Desktop 2. Find `agent0ai/agent-zero` and click the "Run" button 3. Click "Optional settings" to expand configuration options ![Docker run configuration](https://www.agent-zero.ai/res/docs/setup/2-docker-image-run.png) 4. **Port mapping:** In the "Ports" section, enter `0` in the "Host port" field next to port 80 **Tip:** Use `0` to let Docker assign a random port, or specify a port like `50080` 5. **(Optional) SSH access:** Map the SSH port (22) to access the container's terminal 6. **(Optional) Data persistence:** Map volumes to preserve data (see Data Persistence section below) ![Docker port mapping](https://www.agent-zero.ai/res/setup/2-docker-image-run2.png) 8. Click "Run" to start the container #### Option B: Using Command Line # Basic run (ephemeral - data lost on container deletion) docker run -d -p 50080:80 --name agent-zero agent0ai/agent-zero # With data persistence (recommended) docker run -d -p 50080:80 -v /path/to/local/data:/a0/usr --name agent-zero agent0ai/agent-zero The container will start in a few seconds. You'll see a green "Running" status in Docker Desktop: ![Docker container running](https://www.agent-zero.ai/res/docs/setup/4-docker-container-started.png) ### Step 4: Access the Web UI Once the container is running, access Agent Zero through your web browser: * **Via Docker Desktop:** Click on the port link shown in your container details ![Docker container port link](https://www.agent-zero.ai/res/docs/setup/5-docker-click-to-open.png) * **Via Browser:** Navigate to `http://localhost:` (e.g., `http://localhost:50080`) You should see the Agent Zero interface with a green status indicator: ![Agent Zero Web UI](https://www.agent-zero.ai/res/docs/setup/6-docker-a0-running-new.png) **⚠️ Important: Configure Your AI Model** Agent Zero won't work until you configure an AI model provider. Continue to the next section to set this up. How to Configure and Use Agent Zero ----------------------------------- After installation, you need to configure Agent Zero with an AI model provider before you can start chatting. Click the "Settings" button in the web interface to access all configuration options. ### Quick Start: Using OpenRouter OpenRouter is configured as the default provider and requires minimal setup: 1. Visit [openrouter.ai](https://openrouter.ai/) and create an account 2. Navigate to API Keys and create a new key 3. Copy your API key 4. From the 'Models' section of OpenRouter, choose one of the available free models 5. In Agent Zero, click "Settings" → "External Services" 6. Paste your key in "OpenRouter" 7. Write the model names in the 'Chat Model', 'Utility Model' and 'Web Browser Model' fields 8. Click "Save" **✅ Done!** You're ready to start using Agent Zero. Try asking it a question in the chat interface. ### Understanding Model Configuration Agent Zero uses different AI models for different purposes. Each can be configured independently: #### Chat Model (Main Brain) * Primary model that handles conversations and complex reasoning * Configure in: **Settings → Agent Settings → Chat Model** * **Recommended:** gpt-4.1 for quality and cost efficiency ![Chat model settings](https://www.agent-zero.ai/res/setup/2-chat-model.png) #### Utility Model (Helper Model) * Smaller, faster model for internal tasks like memory organization and summarization * Can use the same provider as chat model or a cheaper/faster alternative * **Recommended:** gpt-4.1 for quality and cost efficiency #### Embedding Model (Memory System) * Generates embeddings for Agent Zero's memory and knowledge base * **Note:** Changing this will re-index all memories (can take time) * **Recommended:** Use the default light-weight embedding model ### Security: Setting Up Authentication If you plan to access Agent Zero over the internet or from multiple devices, secure your instance: * **UI Login & Password:** Settings → Authentication → Set username and password for web interface access * **Root Password:** Settings → Authentication → Set password for SSH access to the Docker container After enabling authentication, you can easily share your Agent Zero instance by opening the **External Services** panel in the UI, scrolling to the bottom to find the **Flare Tunnel** section, and clicking **Create Tunnel** to expose it via a shareable QR code. ![Authentication settings](https://www.agent-zero.ai/res/docs/setup/settings/3-auth.png) **⚠️ Important:** Always enable authentication if using Cloudflare Tunnel or exposing your instance to the internet. ### Customizing Agent Behavior Agent Zero's behavior is controlled by prompts and configurations: See the full [Usage Guide](https://github.com/agent0ai/agent-zero/blob/main/docs/usage.md) for detailed customization options. Running Local Models with Ollama -------------------------------- For complete privacy, offline operation, and zero API costs, run local language models using Ollama. **Note:** Local models require significant additional configuration and computing resources. ### Step 1: Install Ollama Download and install Ollama on your **host machine** (not inside Docker): * **Windows/macOS:** Download installer from [ollama.com](https://ollama.com/) * **Linux:** `curl -fsSL https://ollama.com/install.sh | sh` ### Step 2: Download Models Pull the models you want to use. Popular options include: # Recommended for chat ollama pull llama3.2 # Smaller, faster option ollama pull llama3.2:1b # Larger, more capable ollama pull llama3.1:70b View all available models at the [Ollama model library](https://ollama.com/library) . ### Step 3: Configure Agent Zero 1. In Agent Zero, go to **Settings → Agent Settings** 2. For Chat Model (or Utility/Embedding), select **"Ollama"** as the provider 3. Enter the model code (e.g., `llama3.2`) 4. Set API base URL to `http://host.docker.internal:11434` 5. Save your settings ![Ollama local models settings](https://www.agent-zero.ai/res/docs/setup/settings/4-local-models.png) **💡 Note:** The URL `host.docker.internal` allows the Docker container to access services on your host machine. ### Managing Ollama Models # List downloaded models ollama list # Remove a model ollama rm llama3.2 # Update a model ollama pull llama3.2 **⚡ Performance Tip** Local models require significant RAM and CPU/GPU resources. For best performance: * Use models sized appropriately for your hardware (1B-7B for most users) * Consider using a GPU if available * Local models may be slower than cloud APIs, especially for complex tasks **Alternative:** You can also use LM Studio for local models. Configuration is similar to Ollama. Data Persistence & Backups -------------------------- By default, Docker containers are ephemeral. This means that when you delete a container, all data inside is lost. This includes conversation history, memories, files created by agents, and all settings. ### Option 1: Built-in Backup & Restore (Recommended) Agent Zero includes built-in backup functionality: 1. Click **Settings → Backup & Restore** 2. Click "Create Backup" to download all data as a file 3. Store the backup file safely on your computer 4. When updating or recreating containers, use "Restore" to import your data ### Option 2: Volume Mapping (Advanced) Most users do **not** need to map any volumes, using the built-in Backup & Restore is simpler and covers typical everyday usage. Only use volume mapping if you need data to persist across container recreations, want to inspect project files directly from your machine, or want to attach a synced folder (for example a Google Drive or Dropbox folder). When you do need a volume, the **recommended, future-proof way** is to map a folder of your choice on your machine to the `/a0/usr` directory inside the container: docker run -d -p 50080:80 -v /path/to/local/folder:/a0/usr --name agent-zero agent0ai/agent-zero This keeps your projects and other user data stored under `/a0/usr` persistent across container restarts, while Agent Zero's internal system files remain managed by the Docker image so upgrades stay safe. **⚠️ Important Volume Mapping Notes** * Do NOT map the entire `/a0` directory when upgrading. It can cause version conflicts and break your container * Avoid individually mapping internal subdirectories like `/a0/memory`, `/a0/knowledge`, or `/a0/agents` unless you know exactly what you're doing * Prefer a single mapping from a folder on your machine to `/a0/usr`; this is the recommended, future-proof location for projects and user data * Even when using volumes, use the Backup & Restore feature before updates to avoid compatibility issues ### What Gets Saved Agent Zero's data includes: * `/agents` - Agent configurations * `/usr/projects` - Projects * `/memory` - Conversation history and memories * `/knowledge` - Knowledge base files * `/instruments` - Custom tools and instruments * `/prompts` - Custom prompt templates * `/tmp/settings.json` - UI settings and configurations * `.env` - Environment variables and API keys Updating Agent Zero ------------------- Agent Zero is actively developed with regular updates. **Always backup your data before updating!** ### Safe Update Process (Recommended) Keep your current Agent Zero running while setting up the new version. This ensures you can safely backup from the old instance and restore to the new one. 1. **Keep your current container running** - Don't stop or delete it yet! 2. **Pull the latest image:** Download the new Agent Zero version 3. **Run the NEW container on a different port:** `docker run -d -p 50081:80 --name agent-zero-new agent0ai/agent-zero` 4. **Access your OLD instance:** Go to `localhost:50080` 5. **Create a backup:** Settings → Backup & Restore → Create Backup (download the file) 6. **Access your NEW instance:** Go to `localhost:50081` 7. **Restore your backup:** Settings → Backup & Restore → Restore (upload your backup file) 8. **Verify everything works:** Refresh the page and check your chats, projects, and files 9. **Only NOW delete the old container:** Your data is safely migrated! ### Command Line Update Process # Step 1: Pull the latest image (keep old container running!) docker pull agent0ai/agent-zero:latest # Step 2: Run NEW container on different port docker run -d -p 50081:80 --name agent-zero-new agent0ai/agent-zero # Step 3: Create backup from OLD instance (localhost:50080) # Use the UI: Settings → Backup & Restore → Create Backup # Step 4: Restore to NEW instance (localhost:50081) # Use the UI: Settings → Backup & Restore → Restore # Step 5: Verify your data is restored, then remove old container docker stop agent-zero docker rm agent-zero # Step 6: (Optional) Rename new container and switch to original port docker stop agent-zero-new docker rm agent-zero-new docker run -d -p 50080:80 --name agent-zero agent0ai/agent-zero **Why This Process?** Running both containers ensures you can always access your current data. If something goes wrong with the new version, your original Agent Zero is still running with all your data intact. **Never delete your old container until you've verified the new one has all your data!** Mobile & Network Access ----------------------- Access Agent Zero from anywhere on your local network, or expose it to the internet. ### Local Network Access Access Agent Zero from phones, tablets, or other computers on your network: 1. Find your computer's local IP address: * **Windows:** Run `ipconfig` and look for IPv4 Address * **macOS/Linux:** Run `ifconfig` or `ip addr` 2. On your mobile device, navigate to `http://[YOUR_IP]:[PORT]` 3. Example: `http://192.168.1.100:50080` ### Internet Access To access Agent Zero from anywhere via the internet, use the built-in tunnel integration in the UI: 1. In the Agent Zero UI, open **Settings → External Services**. 2. Scroll to the bottom and find the **Flare Tunnel** section. 3. Click **Create Tunnel**. 4. Scan the generated **QR code** on your phone or share the provided URL to access your Agent Zero instance remotely. For advanced or custom configurations, you can also use Cloudflare Tunnel directly; see the official [Cloudflare Tunnel documentation](https://www.cloudflare.com/products/tunnel/) . **⚠️ Security warning:** Always enable authentication (UI login/password) before exposing to the internet! Need Help? ---------- Running into issues? Reach out to the community! ### Community Support Get help from the community and developers in real-time. * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Prompts & Customization - Agent Zero Customization ============= Prompts & Customization ----------------------- Customize Agent Zero's behavior by modifying prompts, creating custom agent profiles, and using dynamic variables for advanced prompt engineering. Understanding the Prompt System ------------------------------- Agent Zero's behavior is controlled through a flexible prompt system using Markdown files. The system supports variable placeholders, dynamic loaders, and agent-specific overrides. ### Core Prompt Files * **agent.system.main.role.md** - Defines the agent's overall role and capabilities * **agent.system.main.communication.md** - Specifies how the agent should communicate * **agent.system.main.solving.md** - Describes the agent's approach to tasks * **agent.system.main.behaviour.md** - Controls dynamic behavior adjustments and rules * **agent.system.main.environment.md** - Defines the runtime environment context * **agent.system.tools.md** - Organizes and calls individual tool prompt files * **agent.system.tool.\*.md** - Individual tool prompt files **💡 Key Concept:** The `agent.system.main.md` file acts as a central hub, referencing other prompt files to create the complete system prompt. Prompt Organization ------------------- ### Default Prompts Located in `/prompts/`, serve as the base configuration: /prompts/ ├── agent.system.main.md ├── agent.system.main.role.md ├── agent.system.main.communication.md ├── agent.system.main.solving.md ├── agent.system.main.behaviour.md ├── agent.system.tools.md └── agent.system.tool.*.md ### Custom Prompts (v0.9.7+) Place overrides in `agents//prompts/`: 1. Create or clone an existing agent profile under `agents//` 2. Add only the prompt files you want to override in `agents//prompts/` 3. Agent Zero merges these overrides with the default prompts automatically 4. Select the **Agent Profile** in Settings to activate the overrides **💡 Benefit:** You only need to override specific prompts while inheriting the rest of the default behavior. Variable Placeholders --------------------- Prompts can include variables using the `{{'{{'}}var{{'}}'}}` syntax. These variables are replaced with actual values when the prompt is processed. ### Example Usage # Current system date and time of user - current datetime: {{'{{'}}date_time{{'}}'}} - rely on this info always up to date # Agent Information - agent name: {{'{{'}}agent_name{{'}}'}} - agent number: {{'{{'}}number{{'}}'}} ### Common Variables * `{{'{{'}}date_time{{'}}'}}` - Current system date and time * `{{'{{'}}agent_name{{'}}'}}` - Name of the current agent * `{{'{{'}}number{{'}}'}}` - Agent number in hierarchy * `{{'{{'}}tools{{'}}'}}` - Available tools list * `{{'{{'}}memory{{'}}'}}` - Retrieved memory context Dynamic Behavior System ----------------------- ### Real-Time Behavior Adjustment Agents can modify their behavior in real-time based on user instructions: * Behavior changes are automatically integrated into the system prompt * Behavioral rules are merged intelligently, avoiding duplicates and conflicts * Changes are applied without disrupting other components **Example:** User: "Please respond in UK English from now on" Agent: Uses behavior_adjustment tool to add this rule Result: All future responses follow UK English conventions ### How It Works 1. User requests behavior changes 2. System identifies behavioral instructions in conversation 3. New rules are merged with existing ruleset 4. Updated behavior is immediately applied 5. Custom rules stored in agent's memory directory as `behaviour.md` **💡 Note:** Behavior rules are injected at the start of the system prompt and maintain separation between core functionality and behavioral rules. Creating Custom Agent Profiles ------------------------------ ### Profile Structure /agents/ └── my_custom_agent/ ├── prompts/ │ ├── agent.system.main.role.md │ ├── agent.system.main.communication.md │ └── agent.system.tool.*.md ├── tools/ │ └── custom_tool.py └── extensions/ └── agent_init/ └── custom_extension.py ### Creating a Profile 1. Create a new directory under `/agents/` 2. Add only the components you want to customize 3. Agent Zero automatically merges with defaults 4. Select your profile in Settings → Agent Configuration **💡 Tip:** Start by cloning the `_example` agent profile and modifying it to your needs. Best Practices -------------- ### Prompt Engineering * **Be specific:** Clear instructions lead to better agent behavior * **Use examples:** Show desired output formats * **Define constraints:** Specify what the agent should and shouldn't do * **Test iteratively:** Adjust prompts based on actual agent responses * **Keep it organized:** Use markdown headers for structure ### Agent Profiles * Create profiles for different use cases (coding, research, automation) * Override only what's necessary - inherit the rest * Document profile purpose and differences from default * Version control your custom profiles ### Maintenance * Keep prompts updated as Agent Zero evolves * Test after framework updates * Balance customization with maintainability * Use Projects for task-specific instructions instead of global prompt changes Need Help? ---------- Questions about prompt customization? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Changelog - v0.6 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6 WebUI & Knowledge System ======================== September 1, 2024 New web UI prototype with Flask server. Knowledge folder for automatic document import. FAISS vector database. Prompt subfolders for easy customization. ### ✨ New Features * Web UI prototype - Flask-based interface with real-time message display and pause/resume controls * Knowledge folder - automatic document import (text, PDF, CSV, HTML, JSON, markdown) into vector database * FAISS vector database - replaced Chroma for better search quality and score threshold support * Prompt subfolders - override default prompts by creating custom subfolders * Separated CLI and UI - run\_cli.py for terminal, run\_ui.py for web interface ### ⚡ Improvements * Real-time message boxes - blue boxes for agent thoughts, tool execution, and responses * JSON visibility - view original LLM JSON output in UI * Background execution - agent runs while UI updates in real-time * Pause and resume controls - interrupt and continue agent execution anytime * initialize.py - centralized configuration for both CLI and UI modes * Prompt override system - only copy files you want to change, not entire default folder * Knowledge subdirectories - organize knowledge base with folder structure * Memory subdirectories - organize memory with custom folder paths * Document splitting - automatic chunking of imported knowledge files ### 🐛 Bug Fixes * Fixed vector database search with score threshold --- # Extensions Framework - Agent Zero Development =========== Extensions Framework -------------------- Agent Zero is built with extensibility in mind, providing a framework for creating custom extensions, tools, API endpoints, and helpers to enhance functionality. Extensibility Overview ---------------------- The Python framework controlling Agent Zero is built as simple as possible, relying on independent smaller and modular scripts for individual tools, API endpoints, system extensions and helper scripts. This way individual components can be easily replaced, upgraded or extended. ### Extensible Components * **Extensions:** Hook into agent lifecycle points to modify behavior * **Tools:** Modular components providing specific functionality to agents * **API Endpoints:** Expose Agent Zero functionality to external systems * **Helpers:** Utility modules providing shared logic across the framework * **Prompts:** Instructions and context provided to the LLM Extensions ---------- Extensions are components that hook into specific points in the agent's lifecycle. They allow you to modify or enhance the behavior of Agent Zero at predefined extension points. The framework uses a plugin-like architecture where extensions are automatically discovered and loaded. ### Extension Points * **agent\_init:** Executed when an agent is initialized * **before\_main\_llm\_call:** Executed before the main LLM call is made * **message\_loop\_start:** Executed at the start of the message processing loop * **message\_loop\_prompts\_before:** Executed before prompts are processed * **message\_loop\_prompts\_after:** Executed after prompts are processed * **message\_loop\_end:** Executed at the end of the message processing loop * **monologue\_start:** Executed at the start of agent monologue * **monologue\_end:** Executed at the end of agent monologue * **reasoning\_stream:** Executed when reasoning stream data is received * **response\_stream:** Executed when response stream data is received * **system\_prompt:** Executed when system prompts are processed ### Extension Mechanism The extension mechanism works through the `call_extensions` function: 1. Loads default extensions from `/python/extensions/{'{{'}}extension_point{{'}}'}/` 2. Loads agent-specific extensions from `/agents/{'{{'}}agent_profile{{'}}'}/extensions/{'{{'}}extension_point{{'}}'}/` 3. Merges them, with agent-specific extensions overriding default ones based on filename 4. Executes each extension in order ### Creating Custom Extensions To create a custom extension: 1. Create a Python class that inherits from the `Extension` base class 2. Implement the `execute` method 3. Place the file in the appropriate extension point directory # File: /agents/_example/extensions/agent_init/_10_example_extension.py from python.helpers.extension import Extension class ExampleExtension(Extension): async def execute(self, **kwargs): # rename the agent to SuperAgent0 self.agent.agent_name = "SuperAgent" + str(self.agent.number) **💡 Override Logic:** When an extension with the same filename exists in both default and agent-specific locations, the agent-specific version takes precedence. Tools ----- Tools are modular components that provide specific functionality to agents. They are invoked by the agent through tool calls in the LLM response. Tools are discovered dynamically and can be extended or overridden. ### Tool Structure Each tool is implemented as a Python class that inherits from the base `Tool` class. Tools are located in: * Default tools: `/python/tools/` * Agent-specific tools: `/agents/{'{{'}}agent_profile{{'}}'}/tools/` ### Tool Override Logic When a tool with the same name is requested, Agent Zero first checks for its existence in the agent-specific tools directory. If found, that version is used. If not found, it falls back to the default tools directory. # File: /agents/_example/tools/response.py from python.helpers.tool import Tool, Response class ResponseTool(Tool): async def execute(self, **kwargs): print("Redefined response tool executed") return Response( message=self.args["text"] if "text" in self.args else self.args["message"], break_loop=True ) ### Tool Execution Flow 1. Tool initialization 2. `before_execution` method 3. `execute` method (main functionality) 4. `after_execution` method API Endpoints ------------- API endpoints expose Agent Zero functionality to external systems or the user interface. They are modular and can be extended or replaced. **Location:** * Default endpoints: `/python/api/` Each endpoint is a separate Python file that handles a specific API request. This modular approach makes it easy to add new endpoints or modify existing ones. Helpers ------- Helper modules provide utility functions and shared logic used across the framework. They support the extensibility of other components by providing common functionality. **Location:** * Default helpers: `/python/helpers/` Prompts ------- Prompts define the instructions and context provided to the LLM. They are highly extensible and can be customized for different agents. ### Prompt Locations * Default prompts: `/prompts/` * Agent-specific prompts: `/agents/{'{{'}}agent_profile{{'}}'}/prompts/` **📝 Note:** Since v0.9.7, custom prompts should be placed under `agents//prompts/` instead of a shared `prompts` subdirectory. ### Prompt Features **Variable Placeholders:** Prompts can include variables using the `{{'{{'}}var{{'}}'}}` syntax: # Current system date and time of user - current datetime: {{'{{'}}date_time{{'}}'}} - rely on this info always up to date **Dynamic Variable Loaders:** Create Python files with the same name as your prompt files to generate variables at runtime: from python.helpers.files import VariablesPlugin from python.helpers import files class MyVariablesPlugin(VariablesPlugin): async def generate(self): # Generate dynamic content tools_list = await self.load_tools() return { "tools": tools_list, "custom_var": "custom value" } **💡 See the [Prompts & Customization](https://www.agent-zero.ai/p/docs/prompts/) guide for detailed information on prompt engineering and customization.** Best Practices -------------- ### Extension Development * Keep extensions focused on a single responsibility * Use appropriate extension points for your use case * Test extensions in isolation before deploying * Document extension behavior and requirements * Use agent-specific extensions for custom agent profiles ### Tool Development * Inherit from the base `Tool` class * Implement clear input validation * Provide meaningful error messages * Use `before_execution` for setup, `after_execution` for cleanup * Test tools with various input scenarios ### Maintenance * Version control your custom extensions and tools * Keep extensions updated with framework changes * Monitor extension performance impact * Review and refactor regularly Need Help? ---------- Questions about extending Agent Zero? Reach out to the community! * [Discord Server](https://discord.gg/B8KZKNsPpj) * [Skool Community](https://www.skool.com/agent-zero) * [GitHub](https://github.com/agent0ai/agent-zero) --- # Changelog - v0.8.5 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.5 MCP Support =========== June 6, 2025 Complete Model Context Protocol (MCP) integration. Agent Zero can now use thousands of external MCP tools and be exposed as an MCP server to other AI applications. ### ✨ New Features * MCP Client - use external MCP tools from third-party developers via JSON configuration * MCP Server - expose Agent Zero as an MCP tool to other AI applications (Cursor, Windsurf, etc.) * NPX/UVX executable module support - run MCP tools on-the-fly without installation * Cloudflare tunnel integration for secure remote MCP server access ### ⚡ Improvements * JSON-based MCP configuration system with custom parameters and descriptions * Token-based MCP authentication generated from username/password * Support for both persistent and ad-hoc chat sessions via MCP * Password protection for tunnel URLs with HTTPS encryption * Remote and local MCP server support (remote like DeepWiki, local in Docker container) * Automatic MCP server initialization and connection status reporting --- # Changelog - v0.8.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.1 Browser Agent & UX ================== January 19, 2025 Browser agent powered by browser-use framework enables autonomous web browsing. Real-time screenshots, interactive element highlighting, and major file handling improvements. ### ✨ New Features * Browser agent - autonomous web browsing using browser-use framework as subordinate agent * Real-time screenshot updates - watch browser agent work with periodic progress reports * Path recognition UI - clickable Linux paths for easy file downloads and navigation * Integrated file browser - open folders from any path segment ### ⚡ Improvements * Browser-use framework integration - 100% LangChain compatible, easy plug-and-play * Optional vision mode - bounding boxes highlight interactive elements on web pages * File attachment handling - upload, compress, merge PDFs, combine images * Clickable file paths - download files directly from terminal output * File conversion capabilities - convert, compress, and sort files without external tools * Privacy-focused file processing - code-based operations keep data local --- # Changelog - v0.7.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.7.1 Memory System & Instruments =========================== October 16, 2024 Complete memory system overhaul with automatic fragment and solution storage. Instruments framework for custom tools. Revamped mobile-friendly UI with light/dark themes. ### ✨ New Features * Automatic memory system - fragments and solutions auto-saved and loaded with metadata filtering * Four memory areas - main, fragments, solutions, instruments with automatic organization * Instruments framework - custom tools executed in Docker container (bash, Python, Node.js scripts) * Knowledge auto-import - markdown files automatically imported to memory from knowledge folder * Reflection prompts - Chain of Thought + reflection for advanced reasoning (Diamantai prompts) * Persistent chats - chat history saved across sessions * Preinstalled binaries - tools ready to use in Docker container ### ⚡ Improvements * Revamped UI - mobile-friendly, responsive, light/dark theme support * Auto-scroll toggle - automatically adjusts based on scroll position * Utility messages grouping - framework messages collapsible with progress indicator * File inclusion system - prompts can include other files with {{include path}} * Extensions framework - modular system for message loop customization * Memory metadata - IDs, timestamps, areas for precise filtering * Similarity-based memory updates - automatic consolidation of related memories * Custom/default knowledge split - personal knowledge ignored by git * Documentation updates and improvements ### 🐛 Bug Fixes * Various bugfixes --- # Changelog - v0.9.5.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.5.1 Venice and xAI Provider Support =============================== September 19, 2025 Added support for Venice AI and xAI providers, expanding model integration options. ### ✨ New Features * Added support for Agent Zero Venice provider * Added support for xAI provider --- # Changelog - v0.8.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.2 Vision & Multitasking ===================== April 5, 2025 Vision support for Claude, Gemini, and GPT models. Terminal sessions enabling true multitasking. Automatic chat renaming with topic tracking. ### ✨ New Features * Vision tool - native image processing for Claude, Gemini, and GPT models with visual information extraction * Terminal sessions - multiple concurrent terminal sessions for parallel execution and multitasking * Automatic chat renaming - utility model-powered topic tracking with background updates ### ⚡ Improvements * Vision toggle in settings per model - enable/disable based on model capabilities * OCR fallback when vision is disabled - extract text from images without vision * Session parameter for code execution - run infinite processes (servers) while using other terminals * Better terminal understanding - agent knows how to manage multiple sessions * Session reset capability - kill processes in specific terminal sessions * Topic change detection - automatic chat rename when conversation shifts --- # Changelog - v0.9.0 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.0 Backup & Subordinate Agent Profiles =================================== June 26, 2025 Starting 0.9 series focused on seamless operation. Complete backup and restore system for easy upgrades. Subordinate agents can now have specialized roles with dedicated prompt profiles. ### ✨ New Features * Backup and restore system - transfer data between instances with configurable folders, dry run support, and conflict resolution * Subordinate agent prompt profiles - agents can select specialized roles (developer, researcher, hacker) for subordinates * Developer and researcher agent prototypes with step-by-step role-specific instructions ### ⚡ Improvements * Prompt override system now per-agent instead of global (only applies to agent zero, subordinates choose their own) * Backup includes chat history, memories, settings, API keys, and custom knowledge * Path translation between Docker and local development environments in backups * Agent can create and modify its own subordinate agent profiles ### 🐛 Bug Fixes * Security fixes --- # Changelog - v0.9.4 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.4 Connectivity & UI Enhancements ============================== August 18, 2025 Major connectivity update with external API endpoints, Agent-to-Agent protocol, new notification system, and enhanced UI/UX improvements. ### ✨ New Features * External API endpoints for third-party application integration * A2A (Agent to Agent) protocol - server and client implementation * New notifications system with front-end and back-end support * Tunnel URL QR code generation for mobile access ### ⚡ Improvements * Streamable HTTP support for MCP server * New local terminal interface for improved stability (replacing SSH) * Rate limiter integration directly into model interfaces * Delayed memory recall option for faster initial responses * Smarter autoscrolling in UI * Action buttons (copy/play) in messages * Multiple API keys support with automatic rotation * Download streaming with progress indication ### 🐛 Bug Fixes * Internal fixes and optimizations --- # Changelog - v0.9.7 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.7 Projects Management System ========================== November 19, 2025 Major release introducing the Projects Management System with per-project custom instructions, integrated memory, knowledge, and files, project-specific secrets, and a new Welcome dashboard and Wait tool for better agent control. ### ✨ New Features * Projects Management System to organize work into dedicated, configurable projects * Support for custom instructions per project to tailor agent behavior to each context * Full integration of projects with memory, knowledge, and file management * Project-specific secrets management to keep credentials isolated and secure * New Welcome screen and dashboard for quick access to projects and tools * New Wait tool to better control and synchronize agent workflows ### ⚡ Improvements * Subordinate agent configuration override support for more flexible orchestration * Support for multiple documents at once in the document query tool * Improved context on interventions for clearer oversight and debugging * OpenRouter embedding support for enhanced vector search and retrieval * Frontend components refactor and polishing for a smoother UI experience * Support for Windows PowerShell in the local TTY utility * More efficient selective streaming for LLMs to reduce latency and token usage * UI output length limit improvements for more stable and readable responses * Update checker to keep Agent Zero up to date with the latest releases ### 🐛 Bug Fixes * SSH metadata output fix for more reliable remote operations * Additional minor fixes and stability improvements across projects and streaming --- # Changelog - v0.9.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.1 LiteLLM Integration & UI Overhaul ================================= July 7, 2025 Complete architectural replacement of Langchain with LiteLLM for unified multi-provider support. Major UI overhaul with collapsible message grouping, improved code output, and mobile-optimized design. ### ✨ New Features * LiteLLM integration - replaced Langchain with unified API for all LLM providers * Reasoning models streaming support with visible chain-of-thought (DeepSeek R1, etc.) * Message grouping system - organized UI with collapsible messages by type * Streamable HTTP MCP servers support * Openrouter as default provider for better flexibility ### ⚡ Improvements * Improved code execution tool output with terminal-like appearance and session info * Scrollable tables and code blocks with enhanced rendering * Mobile-optimized UI with space-efficient design * Auto-collapse/expand for all messages of the same type * LLM API URL field for Azure, Ollama, and custom providers * Unified interface for reasoning tokens across all providers * Enhanced message headings showing relevant context --- # Changelog - v0.9.1.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.1.1 API URL Configuration Fix ========================= July 9, 2025 Fixed LLM API base URL field configuration. ### 🐛 Bug Fixes * Fixed LLM API base URL field configuration --- # Changelog - v0.8.2.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.2.1 JSON Message Fix ================ April 7, 2025 Fixed JSON lead message causing some models to misbehave. ### 🐛 Bug Fixes * Fixed JSON lead message causing some models to misbehave --- # Changelog - v0.6.5 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6.5 Free & Local Models =================== September 13, 2024 Added support for free and local models with bug fixes. ### ✨ New Features * Free and local models support ### 🐛 Bug Fixes * Various bug fixes --- # Changelog - v0.6.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6.1 Terminal & Intervention Fixes ============================= September 6, 2024 Fixed intervention and terminal reset issues. ### 🐛 Bug Fixes * Fixed intervention * Fixed terminal reset --- # Changelog - v0.3 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.3 Memory & GPT-4o-mini ==================== July 19, 2024 Memory enhancements, GPT-4o-mini support, and stability improvements. ### ✨ New Features * GPT-4o-mini support ### ⚡ Improvements * Memory enhancements * Stability improvements --- # Changelog - v0.9.5 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.5 Secrets Management & Progressive Web App ======================================== September 3, 2025 Major update introducing secrets management system allowing agents to use credentials without exposing them to LLMs, enhanced message handling with file references, and progressive web app support. ### ✨ New Features * Secrets management system - agent can use credentials without ever exposing them to LLM providers * Progressive web app (PWA) support ### ⚡ Improvements * Agent can copy/paste messages and files without rewriting them using file references * LiteLLM global configuration field for model parameters * JSON format support in extra model parameters * Custom HTTP headers for browser agent * Tunnel component frontend rework ### 🐛 Bug Fixes * Fixed LLM errors with long IDs by using shorter IDs for files and memories * Fixed timezone change bug * Fixed notifications z-index issue --- # Changelog - v0.5 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.5 Models & Hints ============== August 6, 2024 Added models, bug fixes, and hints improvements. ### ✨ New Features * Added models ### ⚡ Improvements * Hints improvements ### 🐛 Bug Fixes * Various bug fixes --- # Changelog - v0.8.1.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.1.2 Temperature & Reasoning Fix =========================== February 8, 2025 Fixed ChatOpenAI default temperature and reasoning models support. ### 🐛 Bug Fixes * Fixed ChatOpenAI default temperature * Fixed reasoning models support --- # Changelog - v0.8.3.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.3.1 Auto Embeddings & Scheduler =========================== May 6, 2025 Automatic local embedding model for easier onboarding. No external API needed - all-MiniLM-L6 model runs locally on CPU. Automatic vector database reindexing when changing models. ### ✨ New Features * Local embedding model - all-MiniLM-L6 from Hugging Face pre-installed in Docker * Automatic vector database reindexing when changing embedding models ### ⚡ Improvements * CPU-compatible embedding model - no GPU required for memory management * Out-of-the-box operation - no embedding API key needed for providers like Grok and Anthropic * Default embedding switched from OpenAI to local model for new users * Scheduler system updates --- # Changelog - v0.9.3 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.3 Subordinate Agents, Memory & Providers ====================================== August 4, 2025 Major extensibility update allowing subordinate agents to have dedicated prompts, tools, and system extensions. Enhanced memory system with AI consolidation and filtering, plus Venice.ai provider support. ### ✨ New Features * Subordinate agents extensibility framework - dedicated prompts, tools, and system extensions * Venice.ai LLM provider support * Streamable HTTP MCP server support (in addition to server-side events) * Initial agent message for new chats to improve formatting behavior * Docker build support for local images ### ⚡ Improvements * LLM providers now configurable via providers.yaml file * AI memory consolidation when saving memories for improved quality * AI filtering when loading memories for better relevance * Auto memory system configuration controls in settings * Faster startup and restart times with lazy resource loading ### 🐛 Bug Fixes * Fixed file browser issue --- # Changelog - v0.8.6 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.6 Hacking Edition Merge & Browser Upgrade ======================================= June 12, 2025 Merged Agent Zero Hacking Edition into main branch with Kali Linux base. Major browser-use framework upgrade for improved stability. ### ✨ New Features * Merged Hacking Edition into main branch - unified codebase now based on Kali Linux * Hacker prompts subfolder - switchable cybersecurity mode in agent settings * Tunnel provider selection - choose between Serveo (with SSE MCP support) and Cloudflare ### ⚡ Improvements * Browser-use framework upgraded to latest version with enhanced integration * Improved browser agent stability and reliability with more relevant output * New build process with base image and run image for faster builds and frozen software versions * Prompt overriding system with hacker subfolder for cybersecurity tasks --- # Changelog - v0.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.2 Pre-release =========== June 26, 2024 Free JSON format, tool superclass, memory injection, new rate limiter, and agent parameters. ### ✨ New Features * Free JSON format * Tool superclass * Memory injection * New rate limiter * Agent parameters --- # Changelog - v0.8.4.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.4.1 Improvements and Bugfixes ========================= May 21, 2025 Various bugfixes related to context management, message formatting, and scheduler improvements. ### ✨ New Features * New model provider support ### ⚡ Improvements * Message formatting improvements * Scheduler improvements * Compatibility and stability improvements ### 🐛 Bug Fixes * Context management bugfixes * Input tool fix --- # Changelog - v0.8.7 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.7 Markdown Rendering & Document RAG ================================= June 19, 2025 Complete formatting overhaul with markdown, emojis, and LaTeX math notation. Real-time response streaming and new document Q&A tool with vector database support. ### ✨ New Features * Document Q&A tool with RAG - analyze large documents (PDF, Word, etc.) using vector database * Real-time response rendering - immediate streaming of agent responses without waiting for completion * Multi-language document support - analyze documents in any language with automatic translation ### ⚡ Improvements * Markdown formatting in agent responses with emoji icons and LaTeX math notation * Clickable links to files and folders inside Docker container * In-place image rendering in responses * Tables, ordered/unordered lists, and comprehensive report formatting * Vector database integration for handling large documents like entire books --- # Changelog - v0.6.4 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6.4 Event Loop Fix ============== September 13, 2024 Fixed event loops in defer.py. ### 🐛 Bug Fixes * Fixed event loops in defer.py --- # Changelog - v0.9.6 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.6 Memory Management Dashboard =========================== October 2, 2025 Major update introducing the Memory Management Dashboard for transparent control over your agent's memories, plus platform updates to Kali, Python, browser tooling, authentication, and provider integrations. ### ✨ New Features * Memory Management Dashboard to view, search, and edit your AI assistant's memories with full transparency and control * Support for multiple memory folders to isolate knowledge across different projects and prevent context bleed * Context-aware semantic memory search that goes beyond simple keyword matching * Tools to correct, enhance, and delete memories, allowing fine-tuning of your agent's knowledge over time * Export, backup, restore, and transfer complete memory datasets using dedicated utilities * New login screen providing a refreshed onboarding and authentication experience * GitHub Copilot provider support for additional coding assistance options ### ⚡ Improvements * Kali environment update for security and tooling improvements * Python update with dual installation support for more flexible runtime management * Browser Use tool updates for more reliable web automation * LiteLLM retry logic on temporary errors to increase robustness of LLM calls * Overall UX and performance improvements across memory-related workflows ### 🐛 Bug Fixes * Improved stability when managing large memory sets and exporting data * Miscellaneous UI and reliability fixes across dashboards and tools --- # Changelog - v0.7 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.7 Automatic Memory ================ October 6, 2024 Automatic memory system, UI improvements, instruments framework, and reflection prompts. ### ✨ New Features * Automatic memory system * Instruments framework * Reflection prompts ### ⚡ Improvements * UI improvements ### 🐛 Bug Fixes * Various bugfixes --- # Changelog - v0.6.3 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6.3 Bug Fixes ========= September 12, 2024 Multiple bug fixes for Ollama, endpoints, event loops, and error logging. ### 🐛 Bug Fixes * Fixed Ollama context and chat model * Fixed endpoints in .env * Fixed defer.py event loop * Fixed error logging for UI --- # Changelog - v0.8.4.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.4.2 Stability & Hacking Edition Update ================================== May 27, 2025 Major code execution stability improvements with timeout handling and multi-session support. Hacking edition fully synchronized with main branch. Enhanced memory retrieval and faster chat operations. ### ✨ New Features * Dedicated search engine tool - separated from knowledge tool for faster web search access * Multi-session terminal support in hacking edition - run infinite processes in one session while executing commands in another ### ⚡ Improvements * Hacking edition fully up-to-date with main branch - all new tools (vision, scheduler, etc.) now available * Code execution tool stability - timeout and stall detection with 15-second inactivity threshold * Terminal terminator sequence detection for faster control return to agent * System messages explaining timeouts and process state to agent * Memory system precision - more conversational context for better retrieval accuracy * Knowledge tool rework for improved efficiency * Instant chat auto-renaming in background (no longer waits for task completion) * Default models updated to GPT-4o-mini with vision enabled by default * Utility model updated to GPT-4o-mini for speed and cost efficiency * Updated documentation with installation video and reference links * Added OpenRouter leaderboards reference ### 🐛 Bug Fixes * Fixed browser agent reset mechanism when agent hangs or cannot open web pages * Fixed behavior adjustment tool prompt issue --- # Changelog - v0.8.1.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.1.1 DeepSeek R1 Support =================== January 21, 2025 Added support for DeepSeek R1 Reasoner and Chat models - 50x cheaper than O1 with visible reasoning tokens and excellent prompt caching optimization. ### ✨ New Features * DeepSeek R1 Reasoner model support - reasoning model with visible chain-of-thought processing * DeepSeek Chat model support - cost-effective alternative with built-in reasoning capabilities ### ⚡ Improvements * Temperature setting moved to additional arguments - required for Reasoner compatibility * Prompt caching optimization - up to 80% cached tokens for cost efficiency * Reasoning tokens visibility - watch model think through complex problems * Chain of thought built into framework - benefit from reasoning without using Reasoner model --- # Changelog - v0.8.4 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.4 Remote Access & Mobile Support ============================== May 15, 2025 One-click Cloudflare tunnel for accessing Agent Zero from any device over the internet. Mobile-friendly UI with password protection and secure remote access. ### ✨ New Features * Cloudflare tunnel - one-click public URL generation for remote access from mobile and other devices * Password authentication - username and password protection for web UI security * Mobile-friendly UI with expandable input box and optimized layout ### ⚡ Improvements * Single-port forwarding for security - only web UI port exposed, not entire container * Automatic tunnel service running in Docker container * Persistent tunnel across Python process restarts * Login/password warnings before creating public tunnel * Mobile UI enhancements for on-the-go usage --- # Changelog - v0.8.3 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.3 Scheduler System ================ May 4, 2025 Long-awaited scheduler system giving Agent Zero time awareness and future planning. Schedule periodic tasks with Cron, specific dates with Planned tasks, or trigger manually via API. ### ✨ New Features * Cron tasks - periodic execution (every hour, daily, etc.) with cron pattern scheduling * Planned tasks - execution at specific dates and times without periodicity * Manual tasks - ad-hoc execution via agent commands or external API endpoints with tokens * Dedicated scheduler tools - agent can create, modify, delete, and manage tasks autonomously * Task UI in settings - manual task creation and management interface with task table ### ⚡ Improvements * Tasks can run in dedicated chats or continue within existing chat threads * System instructions and attachments support for scheduled tasks * Time awareness - Agent Zero can now plan and execute tasks in the future * API endpoint triggering for external task execution * Full task manipulation toolset - list, delete, modify scheduled tasks --- # Changelog - v0.6.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.6.2 Installation Guide ================== September 8, 2024 Added installation guide and fixed run\_cli.py bug. ### ⚡ Improvements * Added installation guide ### 🐛 Bug Fixes * Fixed run\_cli.py bug --- # Changelog - v0.4 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.4 Bug Fixes & UX ============== July 26, 2024 Bug fixes and UX improvements. ### ⚡ Improvements * UX improvements ### 🐛 Bug Fixes * Various bug fixes --- # Changelog - v0.8 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8 Docker Runtime & Voice Mode =========================== January 10, 2025 Major architecture overhaul - fully Dockerized deployment with one-click setup. Local voice mode with Whisper speech-to-text and browser text-to-speech. SearXNG search engine for private, unlimited web search. ### ✨ New Features * Docker runtime - fully containerized deployment, no more Python/conda setup required * Local speech-to-text - Whisper models (base, small, large, turbo) run in Docker container * Browser text-to-speech - native browser API for low-latency voice output * SearXNG search engine - local, private, unlimited web search aggregating Google, Brave, DuckDuckGo * Knowledge tool - unified search combining web results, memory, and future RAG sources * File browser - integrated UI for browsing Docker container files * Attachments support - upload and work with files in the UI ### ⚡ Improvements * Revamped UI with settings panel and modern design * Context window management system for efficient token usage * Unified environment - agent knows Docker filesystem and capabilities * Pre-installed software in Docker image - complex tools ready out-of-the-box * Voice configuration - silence threshold, duration, language selection * Privacy-focused - all speech processing local, no data sent to OpenAI * Stop speech button - interrupt long agent responses anytime * Docker Desktop support - easy one-click deployment with UI * Port mapping and volume mounting for persistent storage --- # Changelog - v0.9.2 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.2 Speech & Attachments ==================== July 21, 2025 Complete voice interaction with Kokoro text-to-speech and improved Whisper speech-to-text. Enhanced attachment system with better UI and cross-device synchronization. ### ✨ New Features * Kokoro text-to-speech integration - 82M parameter model, CPU-compatible, built into Docker with no setup required * Enhanced speech-to-text with Whisper model - multiple size options, visual feedback, and error-aware transcription * New message attachments system with image viewer and improved cross-device synchronization ### ⚡ Improvements * Optimized streaming system for text-to-speech with chunk-by-chunk synthesis * Smart content parsing - skips tables and code blocks in speech output * Log truncation improvements * Hyperlink target improvements * Component examples * API cleanup --- # Changelog - v0.8.5.1 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.8.5.1 MCP Performance Improvements ============================ June 7, 2025 Improved initialization procedure, faster chat loading, and smoother MCP loading. ### ⚡ Improvements * Improved initialization procedure * Faster chat loading * Longer restart timeout * Smoother MCP loading --- # Changelog - v0.9.8 - Agent Zero [← Back to changelog](https://www.agent-zero.ai/p/docs/changelog/) v0.9.8 Skills, UI Redesign & Git projects ================================== February 2026 One of the largest releases in Agent Zero history. The entire UI has been redesigned with process groups and real-time WebSocket state sync, a new Skills framework replaces the legacy Instruments system, new LLM providers, a message queue, Git-based projects, and dozens of bug fixes round out a release that touches nearly every part of the codebase. ### ✨ New Features * Skills System replacing the legacy Instruments with a new SKILL.md standard for structured, portable agent capabilities. * Built-in skills, and UI support for importing and listing skills * Real-time WebSocket infrastructure replacing the polling-based approach for instant UI state synchronization * Complete UI redesign with process groups to visually group agent actions with expand/collapse support * Timestamps and execution time with tool-specific badges * Step detail modals with key-value and raw JSON display * Collapsible responses with show more/less and copy buttons on code blocks and tables * Message queue system allowing users to queue messages while the agent is still processing * Git-based projects with clone authentication for public and private repositories * In-browser file editor for viewing and editing files without leaving the UI * Welcome screen redesign with info and warning banners for connection security, missing API keys, and system resources * Modular settings architecture - the entire settings UI restructured into dedicated Alpine.js components * Scheduler redesign with standalone modal, separate task list, detail and editor components, and project support * Four new LLM providers: CometAPI, Z.AI, Moonshot AI, and AWS Bedrock * Microsoft Dev Tunnels integration for secure remote access * User data migration to /usr directory for cleaner separation of user and system files * Subagents system with configurable agent profiles for different roles ### ⚡ Improvements * Smooth response rendering and scroll stabilization across chat, terminals, and image viewer * Chat width setting and reworked preferences panel * Image viewer improvements with scroll support and expanded viewer * Redesigned sidebar with reusable dropdown component and streamlined buttons * Inline button confirmations for critical actions * Improved login design and new logout button * File browser enhanced with rename and file actions dropdown * Memory operations offloaded to deferred tasks for better performance * Environment variables can now configure settings via A0\_SET\_\* prefix in .env * Automatic migration with overwrite support for .env, scheduler, knowledge, and instruments directories * Projects support extended to MCP, A2A, and external API * Workdir outside project support for more flexible file organization * Agent number tracking in backend and responses for multi-agent identification ### 🐛 Bug Fixes * Fix file name download issue * Parse LLM chunk fix, Alpine store state save/load * Fix CSRF fstring breaking with double to single quotes * Fix prompt fence stripping policy * Fix GitHub Copilot Vision Requests * Document query QA fix * Fix TTS cleanText function for HTML extraction * Fix TTS streaming race conditions and audio cancellation * Fix API keys saved with correct API\_KEY\_ prefix * Skip empty AI messages in output * Better filename check * Remove hard-coded timeout caps on MCP tool execution * Fix code break in tables * Remove folder restriction in image\_get * Fix autoscroll for chat switching * Fix subordinate grouping, layout alignments * Fix path links in step content * Agent retry on critical error * Fix expanded messages on chat switch * Fix browser screenshot flashing * Fix terminal autoscroll * Skip root password change in non-docker env * Fix allowed hosts, image\_get folders, extensions after termination * Fix upload path from tmp/uploads to usr/uploads * Fix MCP/A2A project activation * Fix scheduler flickering * Proper task cancellation in scheduler, leakage in defer.py * Fix for scheduler and memory dashboard style * Fix editor disappearing in MCP/A2A examples * Fix dropdown behind sidebar-bottom * Fix backend not setting agent number * Fix user-message position * Memory area enum serialization fix * Fix response rendering IDs * Fix backup and file limit * Fix compatibility warning * Remove dummy output message These release notes have been generated by Agent Zero. --- # Changelog - Agent Zero Changelog ========= Release notes and version history for Agent Zero ------------------------------------------------ See what's new in each release and stay updated with the latest features and improvements. February 2026 **v0.9.8** ### [Skills, UI Redesign & Git projects](https://www.agent-zero.ai/p/docs/changelog/0.9.8/) One of the largest releases in Agent Zero history. New Skills framework replacing Instruments, real-time WebSocket state sync, complete UI redesign with process groups, message queue, Git projects, and four new LLM providers. ✨ 17 Features ⚡ 13 Improvements 🐛 37 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.8/) November 19, 2025 **v0.9.7** ### [Projects Management System](https://www.agent-zero.ai/p/docs/changelog/0.9.7/) Major release introducing the Projects Management System with per-project custom instructions, integrated memory, knowledge, and files, project-specific secrets, and a new Welcome dashboard and Wait tool for better agent control. ✨ 6 Features ⚡ 9 Improvements 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.7/) October 2, 2025 **v0.9.6** ### [Memory Management Dashboard](https://www.agent-zero.ai/p/docs/changelog/0.9.6/) Major update introducing the Memory Management Dashboard for transparent control over your agent's memories, plus platform updates to Kali, Python, browser tooling, authentication, and provider integrations. ✨ 7 Features ⚡ 5 Improvements 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.6/) September 19, 2025 **v0.9.5.1** ### [Venice and xAI Provider Support](https://www.agent-zero.ai/p/docs/changelog/0.9.5.1/) Added support for Venice AI and xAI providers, expanding model integration options. ✨ 2 Features [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.5.1/) September 3, 2025 **v0.9.5** ### [Secrets Management & Progressive Web App](https://www.agent-zero.ai/p/docs/changelog/0.9.5/) Major update introducing secrets management system allowing agents to use credentials without exposing them to LLMs, enhanced message handling with file references, and progressive web app support. ✨ 2 Features ⚡ 5 Improvements 🐛 3 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.5/) August 18, 2025 **v0.9.4** ### [Connectivity & UI Enhancements](https://www.agent-zero.ai/p/docs/changelog/0.9.4/) Major connectivity update with external API endpoints, Agent-to-Agent protocol, new notification system, and enhanced UI/UX improvements. ✨ 4 Features ⚡ 8 Improvements 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.4/) August 4, 2025 **v0.9.3** ### [Subordinate Agents, Memory & Providers](https://www.agent-zero.ai/p/docs/changelog/0.9.3/) Major extensibility update allowing subordinate agents to have dedicated prompts, tools, and system extensions. Enhanced memory system with AI consolidation and filtering, plus Venice.ai provider support. ✨ 5 Features ⚡ 5 Improvements 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.3/) July 21, 2025 **v0.9.2** ### [Speech & Attachments](https://www.agent-zero.ai/p/docs/changelog/0.9.2/) Complete voice interaction with Kokoro text-to-speech and improved Whisper speech-to-text. Enhanced attachment system with better UI and cross-device synchronization. ✨ 3 Features ⚡ 6 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.2/) July 9, 2025 **v0.9.1.1** ### [API URL Configuration Fix](https://www.agent-zero.ai/p/docs/changelog/0.9.1.1/) Fixed LLM API base URL field configuration. 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.1.1/) July 7, 2025 **v0.9.1** ### [LiteLLM Integration & UI Overhaul](https://www.agent-zero.ai/p/docs/changelog/0.9.1/) Complete architectural replacement of Langchain with LiteLLM for unified multi-provider support. Major UI overhaul with collapsible message grouping, improved code output, and mobile-optimized design. ✨ 5 Features ⚡ 7 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.1/) June 26, 2025 **v0.9.0** ### [Backup & Subordinate Agent Profiles](https://www.agent-zero.ai/p/docs/changelog/0.9.0/) Starting 0.9 series focused on seamless operation. Complete backup and restore system for easy upgrades. Subordinate agents can now have specialized roles with dedicated prompt profiles. ✨ 3 Features ⚡ 4 Improvements 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.9.0/) June 19, 2025 **v0.8.7** ### [Markdown Rendering & Document RAG](https://www.agent-zero.ai/p/docs/changelog/0.8.7/) Complete formatting overhaul with markdown, emojis, and LaTeX math notation. Real-time response streaming and new document Q&A tool with vector database support. ✨ 3 Features ⚡ 5 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.7/) June 12, 2025 **v0.8.6** ### [Hacking Edition Merge & Browser Upgrade](https://www.agent-zero.ai/p/docs/changelog/0.8.6/) Merged Agent Zero Hacking Edition into main branch with Kali Linux base. Major browser-use framework upgrade for improved stability. ✨ 3 Features ⚡ 4 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.6/) June 7, 2025 **v0.8.5.1** ### [MCP Performance Improvements](https://www.agent-zero.ai/p/docs/changelog/0.8.5.1/) Improved initialization procedure, faster chat loading, and smoother MCP loading. ⚡ 4 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.5.1/) June 6, 2025 **v0.8.5** ### [MCP Support](https://www.agent-zero.ai/p/docs/changelog/0.8.5/) Complete Model Context Protocol (MCP) integration. Agent Zero can now use thousands of external MCP tools and be exposed as an MCP server to other AI applications. ✨ 4 Features ⚡ 6 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.5/) May 27, 2025 **v0.8.4.2** ### [Stability & Hacking Edition Update](https://www.agent-zero.ai/p/docs/changelog/0.8.4.2/) Major code execution stability improvements with timeout handling and multi-session support. Hacking edition fully synchronized with main branch. ✨ 2 Features ⚡ 11 Improvements 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.4.2/) May 21, 2025 **v0.8.4.1** ### [Improvements and Bugfixes](https://www.agent-zero.ai/p/docs/changelog/0.8.4.1/) Various bugfixes related to context management, message formatting, and scheduler improvements. ✨ 1 Feature ⚡ 3 Improvements 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.4.1/) May 15, 2025 **v0.8.4** ### [Remote Access & Mobile Support](https://www.agent-zero.ai/p/docs/changelog/0.8.4/) One-click Cloudflare tunnel for accessing Agent Zero from any device over the internet. Mobile-friendly UI with password protection and secure remote access. ✨ 3 Features ⚡ 5 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.4/) May 6, 2025 **v0.8.3.1** ### [Auto Embeddings & Scheduler](https://www.agent-zero.ai/p/docs/changelog/0.8.3.1/) Automatic local embedding model for easier onboarding. No external API needed - all-MiniLM-L6 model runs locally on CPU. ✨ 2 Features ⚡ 4 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.3.1/) May 4, 2025 **v0.8.3** ### [Scheduler System](https://www.agent-zero.ai/p/docs/changelog/0.8.3/) Long-awaited scheduler system giving Agent Zero time awareness and future planning. Schedule periodic tasks with Cron, specific dates with Planned tasks, or trigger manually via API. ✨ 5 Features ⚡ 5 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.3/) April 7, 2025 **v0.8.2.1** ### [JSON Message Fix](https://www.agent-zero.ai/p/docs/changelog/0.8.2.1/) Fixed JSON lead message causing some models to misbehave. 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.2.1/) April 5, 2025 **v0.8.2** ### [Vision & Multitasking](https://www.agent-zero.ai/p/docs/changelog/0.8.2/) Vision support for Claude, Gemini, and GPT models. Terminal sessions enabling true multitasking. Automatic chat renaming with topic tracking. ✨ 3 Features ⚡ 6 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.2/) February 8, 2025 **v0.8.1.2** ### [Temperature & Reasoning Fix](https://www.agent-zero.ai/p/docs/changelog/0.8.1.2/) Fixed ChatOpenAI default temperature and reasoning models support. 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.1.2/) January 21, 2025 **v0.8.1.1** ### [DeepSeek R1 Support](https://www.agent-zero.ai/p/docs/changelog/0.8.1.1/) Added support for DeepSeek R1 Reasoner and Chat models - 50x cheaper than O1 with visible reasoning tokens and excellent prompt caching optimization. ✨ 2 Features ⚡ 4 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.1.1/) January 19, 2025 **v0.8.1** ### [Browser Agent & UX](https://www.agent-zero.ai/p/docs/changelog/0.8.1/) Browser agent powered by browser-use framework enables autonomous web browsing. Real-time screenshots, interactive element highlighting, and major file handling improvements. ✨ 4 Features ⚡ 6 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8.1/) January 10, 2025 **v0.8** ### [Docker Runtime & Voice Mode](https://www.agent-zero.ai/p/docs/changelog/0.8/) Major architecture overhaul - fully Dockerized deployment with one-click setup. Local voice mode with Whisper speech-to-text and browser text-to-speech. SearXNG search engine for private, unlimited web search. ✨ 7 Features ⚡ 9 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.8/) October 16, 2024 **v0.7.1** ### [Memory System & Instruments](https://www.agent-zero.ai/p/docs/changelog/0.7.1/) Complete memory system overhaul with automatic fragment and solution storage. Instruments framework for custom tools. Revamped mobile-friendly UI with light/dark themes. ✨ 7 Features ⚡ 9 Improvements 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.7.1/) October 6, 2024 **v0.7** ### [Automatic Memory](https://www.agent-zero.ai/p/docs/changelog/0.7/) Automatic memory system, UI improvements, instruments framework, and reflection prompts. ✨ 3 Features ⚡ 1 Improvement 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.7/) September 13, 2024 **v0.6.5** ### [Free & Local Models](https://www.agent-zero.ai/p/docs/changelog/0.6.5/) Added support for free and local models with bug fixes. ✨ 1 Feature 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6.5/) September 13, 2024 **v0.6.4** ### [Event Loop Fix](https://www.agent-zero.ai/p/docs/changelog/0.6.4/) Fixed event loops in defer.py. 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6.4/) September 12, 2024 **v0.6.3** ### [Bug Fixes](https://www.agent-zero.ai/p/docs/changelog/0.6.3/) Multiple bug fixes for Ollama, endpoints, event loops, and error logging. 🐛 4 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6.3/) September 8, 2024 **v0.6.2** ### [Installation Guide](https://www.agent-zero.ai/p/docs/changelog/0.6.2/) Added installation guide and fixed run\_cli.py bug. ⚡ 1 Improvement 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6.2/) September 6, 2024 **v0.6.1** ### [Terminal & Intervention Fixes](https://www.agent-zero.ai/p/docs/changelog/0.6.1/) Fixed intervention and terminal reset issues. 🐛 2 Bug Fixes [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6.1/) September 1, 2024 **v0.6** ### [WebUI & Knowledge System](https://www.agent-zero.ai/p/docs/changelog/0.6/) New web UI prototype with Flask server. Knowledge folder for automatic document import. FAISS vector database. Prompt subfolders for easy customization. ✨ 5 Features ⚡ 9 Improvements 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.6/) August 6, 2024 **v0.5** ### [Models & Hints](https://www.agent-zero.ai/p/docs/changelog/0.5/) Added models, bug fixes, and hints improvements. ✨ 1 Feature ⚡ 1 Improvement 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.5/) July 26, 2024 **v0.4** ### [Bug Fixes & UX](https://www.agent-zero.ai/p/docs/changelog/0.4/) Bug fixes and UX improvements. ⚡ 1 Improvement 🐛 1 Bug Fix [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.4/) July 19, 2024 **v0.3** ### [Memory & GPT-4o-mini](https://www.agent-zero.ai/p/docs/changelog/0.3/) Memory enhancements, GPT-4o-mini support, and stability improvements. ✨ 1 Feature ⚡ 2 Improvements [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.3/) June 26, 2024 **v0.2** ### [Pre-release](https://www.agent-zero.ai/p/docs/changelog/0.2/) Free JSON format, tool superclass, memory injection, new rate limiter, and agent parameters. ✨ 5 Features [Read more →](https://www.agent-zero.ai/p/docs/changelog/0.2/) ---