openclaw/openclaw

Features

• Agent Communication Protocols - A protocol integration allowing external coding harnesses to run as background tasks through a backend plugin.
• Agent Configuration Schemas - Standardized workspace files for defining agent behavior, persona, identity, and memory logs, with automatic processing for specific system files.
• Agent Execution Runtimes - A serialized, end-to-end execution path managing session state, model inference, tool execution, and streaming events.
• Agent Runtimes - A backend component managing the model loop, including prompt reception, tool execution, and turn completion.
• Agent Streaming Interfaces - A streaming interface for tool events, assistant deltas, and lifecycle phases, with support for payload sanitization and duplicate-confirmation suppression.
• Concurrency Managers - A serialization mechanism for agent runs to prevent race conditions, supporting configurable queues for concurrent messaging and tool execution.
• Embedded Agent Runtimes - A runtime environment providing a dedicated workspace, injected persona and memory configurations, and built-in system tools for file operations.
• Memory Management Systems - An optional sub-agent that performs blocking memory recall before reply generation to ensure context-aware responses.
• Session Management Systems - A system providing persistent storage, caching, history limiting, and automatic context compaction for long-running conversations.
• Steering and Streaming Controls - A runtime capability for steering inbound prompts into active runs, supporting message queuing, execution interruption, and block-based streaming output.
• Streaming Response Processors - A processor that chunks reply blocks, strips internal reasoning tags, and extracts directives for media, voice, or reply-to actions.
• Plugin Management Systems - A configuration-driven setup that automatically installs and enables plugins before agent turns begin.
• Runtime Compatibility Contracts - A specification defining runtime responsibilities, including model-loop ownership, thread history management, and support for dynamic tool hooks and context lifecycles.
• Tool Definition Adapters - An adapter that normalizes disparate tool-execution signatures to enable consistent policy filtering, sandbox integration, and cross-provider tool support.
• Agent Configuration Management - A centralized system for managing agent defaults, including workspace paths, skill allowlists, and runtime policies.
• Agent Execution Policies - A resolution policy for determining which harness executes an agent turn, supporting model-scoped, provider-scoped, or automatic resolution.
• Inferred Commitment Memories - Short-lived follow-up memories inferred from conversations, scoped to specific agents and channels, delivered through periodic check-ins.
• Bearer Token Authentication - An API supporting public read access, with write operations requiring a Bearer token.
• Model Provider Authentication - Support for API key and OAuth authentication for model providers, with support for reusing local CLI credentials.
• Token Credential Management - A system for validating token-based credentials, including expiration checks and resolution logic.
• Agent Workspace Environments - A primary working directory for file tools and context, functioning as persistent memory for agent operations.
• AI Coding Tool Adapters - Support for various external harness targets via a backend adapter, enabling integration with multiple AI coding tools.
• MCP Tool Bridges - A bridge allowing CLI backends to consume external tool configurations for interoperability.
• AI Workflow Orchestration Formats - A markdown-based format for orchestrating multi-agent sessions with explicit parallelism, approval-safe execution, and reusable program definitions.
• Agent Tool Registries - A mechanism for plugins to register typed tools that the model can invoke, supporting required and optional tool allowlisting.
• Multi-Agent Routing Systems - A routing mechanism for managing multiple isolated agents, supporting deterministic binding based on channel, account, and identity.
• Agent Configuration Tools - A workspace file for defining agent metadata and avatar paths, applied via identity-setting commands.
• Agent Routing Frameworks - Routing bindings for pinning inbound channel traffic to specific agents, supporting account-level scoping and automatic upgrades.
• Logging Systems - A system providing rolling JSONL logs and console output, with configurable levels, redaction, and diagnostics for transport traffic.
• Background Task Ledgers - A ledger for tracking detached work, including agent runs, sub-agent spawns, and scheduled executions, for inspection and auditing.
• Cloud Infrastructure Deployment Guides - A guide for deploying virtual machines, including network security hardening, bastion host configuration, and automated installation.
• Search Tools - A configurable search tool supporting query parameters, result counts, geographic filtering, language settings, and temporal constraints.
• Backup and Recovery Utilities - A CLI utility for creating portable backups of gateway state, including configuration, authentication, and workspace data.
• CLI Extension Frameworks - A mechanism for plugins to extend the CLI by registering custom command groups and subcommands.
• Command Line Interfaces - A CLI for managing skills and packages, supporting global configuration, proxy settings, authentication, and workspace synchronization.
• Agent Command Line Interfaces - A command-line interface for executing agent turns, supporting session management, local runtime overrides, and automated reply delivery.
• Slash Command Interfaces - A user-invocable command that routes to virtual machine instructions and leverages underlying tools for managing AI workflow programs.
• Quality Gates - A suite of commands for type checking and linting to ensure code quality during development.
• Agent Protocol Bridges - A bridge utility enabling IDEs to communicate with a gateway using a standard protocol, supporting session routing and streaming updates.
• Chat Command Interfaces - A system supporting native platform commands, text-based slash commands, and plugin-specific command registration.
• CLI Plugin Architectures - Plugins allowing the integration of local command-line AI tools as inference providers, enabling custom execution and fallback capabilities.
• Debugging Workflows - A manual testing workflow involving gateway development mode, direct agent triggering, and an interactive terminal interface.
• Update Preview Utilities - A utility to preview update operations, inspecting the target version and planned actions before applying changes.
• Version Inspection Tools - A CLI command to inspect the active update channel, installation method, version, and configuration source.
• Workflow State Backends - Support for multiple state backends, including filesystem, transient storage, and database backends for managing workflow state.
• Maintenance Automation Agents - Automated maintenance workflows using event-driven agents to keep documentation aligned with code and perform test fixes.
• Patch Management Tools - A tool enabling agents to perform multi-file or multi-hunk edits, including additions, updates, and deletions, using a specialized patch format.
• Release Channel Managers - CLI support for multiple update channels, allowing users to persist their preferred release stream via configuration.
• CLI Reporting Tools - A CLI feature providing structured JSON metadata for message delivery outcomes to enable programmatic handling in scripts.
• Hook Discovery Mechanisms - A multi-layered mechanism for loading hooks from bundled, plugin-managed, user-defined, and workspace-specific directories with defined override precedence.
• Standing Order Authorities - Persistent operating authority injected into agent sessions, allowing for defined programs and time-based enforcement.
• Cron Schedulers - A scheduler that persists jobs, manages execution state, and supports delivery via chat channels or webhooks.
• Task Schedulers - A scheduler for precise timing that persists jobs, wakes agents, and delivers output to chat channels or webhook endpoints.
• Smart CI Pipelines - A CI pipeline that classifies diffs to skip unnecessary jobs, uses fail-fast ordering, and supports manual dispatches for validation.
• Rate Limiters - A system enforcing rate limits based on IP or user identity, providing standard headers for monitoring and retry logic.
• Automated Infrastructure Deployment - An automated installer for configuring firewalls, VPN access, containerized sandboxes, and system services.
• Background Session Managers - A management interface for background sessions, supporting listing, polling, input delivery, and process termination.
• Configuration-Only Auth Routing - A mechanism for defining routing metadata in configuration files, separating infrastructure routing from credential management.
• Durable Task Orchestrators - A durable orchestration substrate for managing multi-step flows with revision tracking, sync modes, and inspection capabilities.
• Automation Hooks - A collection of pre-built hooks for session memory persistence, command logging, and compaction notifications, configurable via gateway settings.
• Desktop Automation Plugins - A plugin that enables local desktop control by preparing the application server and managing native tool calls.
• Virtual Private Server Deployment Guides - A guide for deploying the gateway on a virtual private server, including system bootstrap and service installation.
• Heartbeat Monitors - A monitoring mechanism that batches periodic checks for inbox, calendar, and notifications into a single scheduled turn.
• Configuration Management Tools - A utility for bulk-replacing approval configurations using structured data files.
• Container Orchestration Helpers - A shell-helper for managing containerized gateway instances, including lifecycle operations and configuration maintenance.
• Agent Testing Frameworks - A test suite for validating agent logic, tool definitions, and live provider interactions.
• Search Engine Integrations - A web search integration supporting standard results and context-grounded text chunks for AI applications.
• Messaging Systems - A system for managing multi-channel messaging integrations, including access policies, model overrides, and platform-specific settings.
• Communication Gateways - A long-running process managing WebSocket control planes, loopback connections, and remote access tunnels.
• Service Discovery and Transport Layers - A layer supporting multiple connectivity methods, including mDNS, SSH tunneling, and secure mesh networking.
• Service Discovery Protocols - A plugin for local network service advertisement and wide-area discovery over secure tunnels.
• Chat Moderation Tools - A gating mechanism for group chats that enforces mention requirements and controls reply behavior.
• Chat Platform Integrations - A capability for redirecting agent replies to different chat platforms while maintaining conversation context and history.
• Message Handling Systems - A system supporting response prefixes, acknowledgement reactions, inbound debouncing, and text-to-speech integration.
• Email Systems - A system for triggering automated workflows from inbox events using a message-queue service, including watch management and model overrides.
• Child Process Management Helpers - A utility for bridging child processes to ensure termination signals are forwarded and listeners are detached.
• Acceptable Usage Policies - A policy prohibiting security-bypass, platform abuse, fraud, scraping, impersonation, and unsafe execution workflows.
• Authentication - A utility for invalidating API tokens, requiring re-authentication to generate new credentials.
• Web Sign-in Flows - A web-based authentication flow for managing accounts and accessing services.
• Credential Management Policies - A policy allowing agents to resolve credentials from a central store without duplication, with specific rules per credential type.
• Device Pairing Protocols - A device-based pairing system requiring explicit approval, utilizing signed nonces and device tokens for secure authentication.
• Identity and Access Management Systems - A system for managing secure device authentication, token rotation, and approval workflows.
• Session Management - A mechanism to bind conversations to specific sessions, allowing transport, authentication, and delivery management within a defined workspace.
• Access Control - A system for granular control over credential usage through chat commands and agent-specific overrides.
• Allowlist Management Utilities - CLI helpers for managing allowlists, enabling users to control executable paths and agent-node associations.
• Execution Policy Managers - A command for synchronizing execution policy configurations and host approval files, supporting presets and inspection.
• Plugin Architectures - A pattern for integrating new capabilities by defining core contracts, extending the plugin API, and wiring implementations through a registry.
• Daemon Architectures - A central daemon managing provider connections, exposing a typed WebSocket API, and validating frames against schemas.
• Capability Ownership Models - An ownership model where plugins act as boundaries, consuming shared capability contracts rather than implementing ad hoc integrations.
• Plugin Format Adapters - A mapping system that converts external plugin formats into native skills, hooks, and tools.
• Plugin Management Interfaces - Chat-surface commands for managing plugin installation, marketplace configuration, and status verification.
• Lifecycle Event Hooks - Event-driven scripts triggered by agent lifecycle events, session management, or message flow, managed through a command interface.
• Credential Storage Utilities - A system for storing API tokens in platform-specific configuration files with support for custom path overrides.
• Event Handling Systems - An event-driven architecture allowing hooks to listen for gateway actions, including command execution, session compaction, and agent bootstrapping.
• Browser Control APIs - A loopback HTTP API for controlling browser instances, managing tabs, executing navigation, and configuring browser state.
• Browser Session Managers - An integration allowing agents to attach to and control an existing, authenticated browser session.
• Browser Snapshotting Systems - A system that generates actionable references for browser elements using numeric refs, role-based selectors, or tree nodes for interaction.
• Managed Browser Environments - A dedicated, isolated browser profile managed by the gateway for performing automated tasks without affecting personal sessions.
• Backend Development - Hooks for fine-grained control over configuration normalization, execution arguments, system prompt transformations, and authentication profiles.
• Browser Security Policies - A security policy configuration for browser instances that allows restricting network access, blocking destinations, and managing script execution permissions.
• Metadata APIs - Read-only endpoints for cataloging and searching skill metadata, requiring responsible consumption via caching.
• Runtime Management & Utilities - A typed API allowing plugins to access core functionality like text-to-speech and media understanding.
• Runtimes & Execution Environments - A plugin-based runtime adapter for integrating external agent harnesses, supporting custom command paths, versioning, and dependency management.
• Diagnostics - Commands for verifying model authentication status and troubleshooting credential issues.
• Security Guardrails - A security layer preventing the use of static secret references for OAuth profiles to ensure credential integrity.
• Session Management Utilities - Automated memory-buffer compaction and summary management, with retry logic to maintain session history and output quality.
• Background Execution Tools - A shell execution tool supporting backgrounding, automatic timeouts, sandbox execution, and TTY allocation.
• State Reset Utilities - A procedure for clearing configuration, authentication, session history, and workspace state.
• Browser Troubleshooting Guides - A diagnostic guide for resolving browser-launch failures on Linux, addressing sandboxing and environment-variable configuration.