Feature Atlas

A structured map of every OpenClaw feature — organized by category, cross-referenced to guide sections, and annotated with known issues. Use this page to understand what’s available, where to configure it, and what to watch out for.

Feature Overview

  %%{init: { "theme": "base", "themeVariables": { "fontFamily": "ui-sans-serif, -apple-system, BlinkMacSystemFont, Segoe UI, Helvetica, Arial", "primaryColor": "#1F4E79", "primaryBorderColor": "#93C5FD", "primaryTextColor": "#F8FAFC", "lineColor": "#94A3B8", "secondaryColor": "#334155", "tertiaryColor": "#334155", "background": "transparent" } }}%%
graph TB
    subgraph CH["Channels & Messaging"]
        direction TB
        CH1["WhatsApp · Signal · Google Chat"]
        CH2["DM & Group Policies"]
        CH3["Mention Gating · Chat Commands"]
    end

    subgraph AG["Agents & Configuration"]
        direction TB
        AG1["Agent Definitions · Routing"]
        AG2["A2A · Subagents · LLM Providers"]
        AG3["Config Includes · Validation"]
    end

    subgraph SM["Sessions & Memory"]
        direction TB
        SM1["Session Scoping (4 modes)"]
        SM2["Two-Layer Memory · Hybrid Search"]
        SM3["Compaction · Temporal Decay"]
    end

    subgraph SE["Security & Hardening"]
        direction TB
        SE1["Sandbox Modes · Docker / VM"]
        SE2["Tool Policies (8 layers)"]
        SE3["5 Guard Plugins · SOUL.md"]
    end

    subgraph TA["Tools & Automation"]
        direction TB
        TA1["44 Tools in 8 Groups"]
        TA2["Cron Jobs · Web Search"]
        TA3["Browser · Computer Use · Image Gen"]
    end

    subgraph DO["Deployment & Operations"]
        direction TB
        DO1["Gateway · LaunchAgent · systemd"]
        DO2["Docker / VM Deploy · Tailscale"]
        DO3["Health Endpoints · Multi-Gateway"]
    end

    subgraph IN["Plugin System & Internals"]
        direction LR
        IN1["Plugin Hooks (6)"]
        IN2["Session Router"]
        IN3["HTTP API · Control UI"]
    end

    CH -->|connects| AG
    AG -->|uses| SM
    AG -->|invokes| TA
    SE -.->|enforces| TA
    TA -->|runs on| DO
    TA -.->|built on| IN

    classDef channels fill:#15803d,stroke:#86EFAC,color:#F8FAFC,stroke-width:1.5px
    classDef agents fill:#1d4ed8,stroke:#93C5FD,color:#F8FAFC,stroke-width:1.5px
    classDef sessions fill:#7c3aed,stroke:#C4B5FD,color:#F8FAFC,stroke-width:1.5px
    classDef security fill:#dc2626,stroke:#FCA5A5,color:#F8FAFC,stroke-width:1.5px
    classDef tools fill:#0891b2,stroke:#7DD3FC,color:#F8FAFC,stroke-width:1.5px
    classDef deploy fill:#d97706,stroke:#FCD34D,color:#F8FAFC,stroke-width:1.5px
    classDef internals fill:#334155,stroke:#94A3B8,color:#F8FAFC,stroke-width:1px

    class CH1,CH2,CH3 channels
    class AG1,AG2,AG3 agents
    class SM1,SM2,SM3 sessions
    class SE1,SE2,SE3 security
    class TA1,TA2,TA3 tools
    class DO1,DO2,DO3 deploy
    class IN1,IN2,IN3 internals

Editable source: The diagram is also available as an Excalidraw file — open it at excalidraw.com for a richer, editable version.

Agents & Configuration

How agents are defined, routed, and connected to each other.

Features

Use Cases

• Single agent with search delegation — main agent + isolated search agent (Phase 5 )
• Multi-channel routing — separate agents per channel with binding patterns (Phase 4 )
• Workspace isolation — per-agent SOUL.md to enforce different behavioral rules (Phase 3 )
• Config splitting — $include for managing complex multi-agent configs (Reference )

Known Issues

Channels & Messaging

How external users communicate with agents through messaging platforms.

Features

Use Cases

• WhatsApp personal assistant — pairing-based DMs with group mention gating (Phase 4 )
• Signal secure messaging — privacy-focused channel with regex mention patterns (Phase 4 , Phase 6 )
• Google Chat workspace bot — GCP service account for team/org use (Google Chat )
• Morning briefing delivery — cron job with delivery.to for automated group reports (Morning Briefing recipe )
• Multi-channel routing — different agents respond on different channels (Phase 4 )

Known Issues

Sessions & Memory

How conversations are scoped, persisted, and how agents remember across sessions.

Features

Use Cases

• Personal assistant with long-term recall — hybrid search with temporal decay (Phase 2 )
• Knowledge vault — structured memory for research and reference material (Knowledge Vault recipe )
• Privacy-first memory — local embeddings, no external API calls (Phase 2 )
• Multi-agent shared context — agents in the same workspace share memory files (Phase 4 )

Known Issues

No major open issues affecting sessions or memory.

Security & Hardening

Layers of protection from sandbox isolation to network controls.

Features

Use Cases

• Pragmatic single agent — no Docker, guard plugins as the safety net (Pragmatic Single Agent )
• Hardened multi-agent — Docker sandbox + all five guard plugins + network egress (Hardened Multi-Agent )
• VM-based isolation — strongest host separation for high-security deployments (Phase 6 )
• Search agent isolation — content-guard scans search results before they reach the main agent (Phase 5 )
• Security audit — automated posture check with openclaw security audit --deep (Security Audit example )

Known Issues

Tools & Automation

The 44 built-in tools, cron scheduling, web search, browser, and extended capabilities.

Features

Use Cases

• Isolated web search agent — web_search + web_fetch on a dedicated agent, denied on main (Phase 5 )
• Automated morning briefing — cron job with isolated session + announce delivery (Morning Briefing recipe )
• Knowledge vault management — filesystem tools + memory for structured research (Knowledge Vault recipe )
• Image generation in chat — generate_image tool via image-gen plugin (Extension )
• VM computer automation — execute tasks in a macOS VM via computer-use plugin (Phase 8 )

Known Issues

Deployment & Operations

Running OpenClaw in production: service management, infrastructure, and day-to-day operations.

Features

Use Cases

• Single-machine production — LaunchAgent + Docker sandbox + Tailscale (Phase 6 )
• Hardened daemon — LaunchDaemon with dedicated user, no login shell (Phase 6 )
• Multi-user setup — separate OS users per gateway for channel separation (Multi-Gateway )
• Machine migration — step-by-step procedure for moving everything to new hardware (Phase 7 )
• Custom tooling — pre-built sandbox images with project-specific packages (Custom Sandbox Images )

Known Issues

No major open issues affecting deployment.

Internals

Plugin System & Architecture

How the gateway works under the hood — the module system, plugin lifecycle, and extension points.

Features

Use Cases

• Custom guard plugin — use before_tool_call hook to intercept and validate tool calls (Extension docs )
• Input/output logging — llm_input/llm_output hooks for compliance and auditing (Reference )
• Custom tool plugin — register new tools accessible by agents (e.g., image-gen, computer-use) (Extension docs )
• API integration — use the HTTP API for programmatic agent interaction (Architecture )

Known Issues