# RadarOS

## Docs

- [A2A Client](https://docs.xhipai.com/a2a/client.md): Connect to remote A2A agents — use them as tools, team members, or call them directly.
- [API Gateway](https://docs.xhipai.com/a2a/gateway.md): Merge local and remote endpoints under a single API surface
- [A2A Protocol](https://docs.xhipai.com/a2a/overview.md): Agent-to-Agent (A2A) protocol support — expose your agents as A2A servers and call remote A2A agents.
- [Remote Teams](https://docs.xhipai.com/a2a/remote-team.md): Call remote team endpoints as local team objects
- [Remote Workflows](https://docs.xhipai.com/a2a/remote-workflow.md): Execute remote workflow endpoints
- [A2A Server](https://docs.xhipai.com/a2a/server.md): Expose RadarOS agents as A2A-compliant HTTP endpoints for cross-framework interoperability.
- [Admin REST API](https://docs.xhipai.com/admin/express.md): CRUD endpoints for managing agents, teams, and workflows via Express.
- [Admin API](https://docs.xhipai.com/admin/overview.md): Dynamically create, update, and delete agents, teams, and workflows at runtime via REST or Socket.IO.
- [Admin Socket.IO](https://docs.xhipai.com/admin/socketio.md): Real-time CRUD events for managing agents, teams, and workflows via Socket.IO with live broadcasts.
- [Admin Storage](https://docs.xhipai.com/admin/storage.md): Configure persistence backends for admin-created agents, teams, and workflows.
- [Multi-Tenant Isolation](https://docs.xhipai.com/advanced/multi-tenant.md): Tenant-level data isolation for SaaS products with namespace or strict separation
- [Rate Limiting & Backpressure](https://docs.xhipai.com/advanced/rate-limiting.md): Token-aware rate limiting with per-tenant scoping and graceful degradation
- [Human-in-the-Loop](https://docs.xhipai.com/agents/approval.md): Require human approval before executing sensitive tool calls. Supports CLI callbacks, event-driven approval via Socket.IO, and configurable timeouts.
- [Run Cancellation](https://docs.xhipai.com/agents/cancellation.md): Abort running agents mid-execution with AbortSignal
- [Context Pollution Prevention](https://docs.xhipai.com/agents/context-curation.md): Automatically curate conversation context to prevent drift from failed results and stale messages
- [Dependencies & Runtime Injection](https://docs.xhipai.com/agents/dependencies.md): Inject runtime variables into agent instructions and messages
- [Followup Suggestions](https://docs.xhipai.com/agents/followups.md): Auto-generate actionable followup prompts after agent responses
- [Hooks & Guardrails](https://docs.xhipai.com/agents/hooks-and-guardrails.md): AgentHooks lifecycle, input/output guardrails, and GuardrailResult in RadarOS.
- [Logging](https://docs.xhipai.com/agents/logging.md): Logger class, LogLevel, AgentConfig logLevel, and formatted agent run output in RadarOS.
- [Memory](https://docs.xhipai.com/agents/memory.md): Complete guide to RadarOS unified memory — sessions, summaries, user facts, profiles, entities, decisions, graph, procedures, and context budgeting.
- [Multi-Modal Input](https://docs.xhipai.com/agents/multimodal.md): Send images, audio, and files to agents. MessageContent, ContentPart types, and examples.
- [Agents](https://docs.xhipai.com/agents/overview.md): Core agent concepts, AgentConfig, run/stream methods, and RunOutput in RadarOS.
- [Streaming Progress Protocol](https://docs.xhipai.com/agents/progress.md): Rich progress events for real-time visibility during long-running agent tasks
- [Reasoning](https://docs.xhipai.com/agents/reasoning.md): Enable extended thinking and reasoning capabilities for agents using OpenAI, Anthropic, Google, and Vertex AI models.
- [Agent Reflection & Self-Correction](https://docs.xhipai.com/agents/reflection.md): Enable agents to critique their own outputs, detect loops, and learn from failures
- [Sandbox Execution](https://docs.xhipai.com/agents/sandbox.md): Run tools in isolated subprocesses with timeout, memory limits, and restricted filesystem/network access.
- [Agent Serialization](https://docs.xhipai.com/agents/serialization.md): Save and load agent configurations to/from JSON
- [Sessions](https://docs.xhipai.com/agents/sessions.md): Session management for multi-turn conversations. Message history, state, overflow handling, and storage drivers.
- [Structured Output](https://docs.xhipai.com/agents/structured-output.md): Enforce JSON schema on LLM responses with Zod. RunOutput.structured and provider support.
- [Tool Caching](https://docs.xhipai.com/agents/tool-caching.md): Cache tool execution results to avoid redundant API calls and speed up agent runs.
- [Tools & Function Calling](https://docs.xhipai.com/agents/tools.md): Define tools with defineTool(), Zod schemas, and execute function calling in RadarOS agents.
- [Types Reference](https://docs.xhipai.com/agents/types-reference.md): Complete reference for every TypeScript interface used across RadarOS — RunContext, ChatMessage, ModelResponse, StreamChunk, TokenUsage, and more.
- [User Memory](https://docs.xhipai.com/agents/user-memory.md): Cross-session personalization with automatic fact extraction and persistent user profiles.
- [Agent Versioning & A/B Testing](https://docs.xhipai.com/agents/versioning.md): Version agent configurations and safely test changes on live traffic
- [@radaros/core](https://docs.xhipai.com/api-reference/core.md): Complete API reference for the core package
- [@radaros/queue](https://docs.xhipai.com/api-reference/queue.md): Complete API reference for the queue package
- [@radaros/transport](https://docs.xhipai.com/api-reference/transport.md): Complete API reference for the transport package
- [Browser Agents](https://docs.xhipai.com/browser/overview.md): Vision-based autonomous browser automation. An agent sees screenshots, decides actions, and operates web pages via Playwright.
- [Semantic Cache](https://docs.xhipai.com/cache/overview.md): Cache LLM responses by semantic similarity to reduce costs and latency
- [Capacity Planner](https://docs.xhipai.com/capacity/capacity-planner.md): Plan GPU requirements, session capacity, and infrastructure sizing for LLM deployments
- [KV Estimator](https://docs.xhipai.com/capacity/kv-estimator.md): Calculate KV cache sizes, memory requirements, and weight memory for any model architecture
- [Latency Estimator](https://docs.xhipai.com/capacity/latency-estimator.md): Model TPOT, TTFT, prefill time, and NAND restore latency for LLM inference
- [Capacity Planning](https://docs.xhipai.com/capacity/overview.md): Estimate GPU requirements, KV cache sizing, latency, and infrastructure cost for self-hosted LLM deployments
- [Session Profiler](https://docs.xhipai.com/capacity/session-profiler.md): Runtime session monitoring with automatic classification and KV cache estimation
- [Compliance & Audit Trail](https://docs.xhipai.com/compliance/overview.md): Tamper-evident audit logging for EU AI Act compliance and enterprise governance
- [Data Retention & Erasure](https://docs.xhipai.com/compliance/retention.md): GDPR-compliant data retention policies and right-to-erasure support
- [Cost Tracking](https://docs.xhipai.com/cost/overview.md): Track token usage, compute costs, and enforce budgets across all agent types
- [Culture System](https://docs.xhipai.com/culture/overview.md): Shared organizational knowledge that compounds over time
- [Bluetooth BLE](https://docs.xhipai.com/edge/ble.md): Scan, connect, and communicate with Bluetooth Low Energy devices.
- [Camera](https://docs.xhipai.com/edge/camera.md): Capture photos and video from the Raspberry Pi camera module.
- [Edge-Cloud Sync](https://docs.xhipai.com/edge/edge-cloud.md): Heartbeat, config pull, event push, and offline-first queue for edge-cloud hybrid architectures.
- [GPIO](https://docs.xhipai.com/edge/gpio.md): Control Raspberry Pi GPIO pins — read, write, watch edges, and software PWM.
- [Ollama on Edge](https://docs.xhipai.com/edge/ollama-edge.md): Model recommendations, auto-start, and pull helpers for running Ollama on Raspberry Pi.
- [Edge & IoT](https://docs.xhipai.com/edge/overview.md): Deploy AI agents on Raspberry Pi and edge devices with IoT toolkits, local LLM inference, and cloud sync.
- [Edge Runtime](https://docs.xhipai.com/edge/runtime.md): Watchdog, resource monitoring, health checks, and graceful degradation for edge devices.
- [I2C Sensors](https://docs.xhipai.com/edge/sensor.md): Read temperature, humidity, and pressure from I2C sensors like BME280.
- [Accuracy Evaluation](https://docs.xhipai.com/eval/accuracy.md): Measure how accurately your agent answers questions
- [Agent Judge Evaluation](https://docs.xhipai.com/eval/agent-judge.md): Custom quality criteria evaluation with LLM judges
- [Conversational Testing](https://docs.xhipai.com/eval/conversational.md): Test multi-turn agent conversations with synthetic users and trajectory scoring
- [Eval Framework](https://docs.xhipai.com/eval/overview.md): Automated agent quality testing with configurable scorers
- [Performance Evaluation](https://docs.xhipai.com/eval/performance.md): Measure agent latency, token usage, and memory consumption
- [Reliability Evaluation](https://docs.xhipai.com/eval/reliability.md): Verify tool usage, error handling, and response completeness
- [Advanced Features Examples](https://docs.xhipai.com/examples/advanced-features.md): Loop hooks, context compaction, approval gates, PII guard, checkpointing, and more
- [Agent Lifecycle & Advanced](https://docs.xhipai.com/examples/agent-lifecycle.md): Hooks, cancellation, follow-up suggestions, serialization, checkpointing, dynamic tools, event bus, context compaction, dependencies, and compression examples.
- [Agent Examples](https://docs.xhipai.com/examples/agents.md): Complete code examples for creating, configuring, and running agents
- [All Model Providers](https://docs.xhipai.com/examples/all-providers.md): Complete code examples for every supported LLM provider in RadarOS.
- [Capacity Planning](https://docs.xhipai.com/examples/capacity-planning.md): Examples for KV cache sizing, GPU planning, latency modeling, and live monitoring
- [Cost & Observability Examples](https://docs.xhipai.com/examples/cost-observability.md): Cost tracking, budget enforcement, tracing, metrics, and structured logging
- [Edge & IoT Examples](https://docs.xhipai.com/examples/edge-iot.md): Deploy agents on Raspberry Pi with GPIO, sensors, camera, and BLE
- [End-to-End Workflow Examples](https://docs.xhipai.com/examples/end-to-end.md): Complete application examples combining agents, teams, workflows, memory, tools, and deployment
- [Integration Examples](https://docs.xhipai.com/examples/integrations.md): MCP, A2A, webhooks, queues, scheduling, and admin management
- [Knowledge & RAG Examples](https://docs.xhipai.com/examples/knowledge-rag.md): Retrieval-augmented generation patterns with vector stores and embeddings
- [Memory & Sessions Examples](https://docs.xhipai.com/examples/memory-sessions.md): Examples for sessions, memory, and user personalization
- [Model Provider Examples](https://docs.xhipai.com/examples/models.md): Configuration examples for every supported LLM provider
- [Production Features](https://docs.xhipai.com/examples/production-features.md): Examples for model resilience, agent reflection, versioning, compliance, multi-tenant, rate limiting, scheduling, conversational testing, context curation, and streaming progress
- [Security & Authentication](https://docs.xhipai.com/examples/security-auth.md): JWT authentication, API key management, PII guard, input/output guardrails, sandbox execution, and human-in-the-loop approval examples
- [Storage Examples](https://docs.xhipai.com/examples/storage.md): Storage driver configuration for every persistence backend
- [Team Examples](https://docs.xhipai.com/examples/teams.md): Multi-agent team patterns: coordinate, route, broadcast, and collaborate
- [All Toolkits](https://docs.xhipai.com/examples/toolkits-extended.md): Complete examples for every toolkit available in RadarOS, showing how to import, configure, and use each one with an agent.
- [Tools & Toolkits Examples](https://docs.xhipai.com/examples/tools-toolkits.md): Custom tools, built-in toolkits, and advanced tool patterns
- [Transport & Deployment Examples](https://docs.xhipai.com/examples/transport-deploy.md): Serve agents via REST APIs, WebSockets, Swagger UI, and file uploads
- [Voice & Browser Agent Examples](https://docs.xhipai.com/examples/voice-browser.md): Real-time voice conversations and autonomous browser automation
- [Workflow Examples](https://docs.xhipai.com/examples/workflows.md): Step-based workflow patterns: linear, conditional, parallel, and complex pipelines
- [Approval Gates](https://docs.xhipai.com/features/approval-gates.md): Human-in-the-loop approval for tool calls
- [Checkpointing](https://docs.xhipai.com/features/checkpointing.md): Agent checkpointing and rollback for fault tolerance
- [Context Compression](https://docs.xhipai.com/features/compression.md): Auto-compress verbose tool results to stay within context windows
- [Context Compaction](https://docs.xhipai.com/features/context-compaction.md): Token-aware context auto-compaction to prevent context window overflow
- [Cost Auto-Stop](https://docs.xhipai.com/features/cost-autostop.md): Per-roundtrip budget enforcement to prevent runaway costs
- [Discovery Cards](https://docs.xhipai.com/features/discovery-cards.md): A2A-style agent discovery cards for inter-agent communication
- [Dynamic Tools](https://docs.xhipai.com/features/dynamic-tools.md): Add, remove, and resolve tools at runtime
- [Loop Hooks](https://docs.xhipai.com/features/loop-hooks.md): Per-roundtrip hooks for fine-grained LLM loop control
- [Metrics Export](https://docs.xhipai.com/features/metrics-export.md): Real-time agent metrics for dashboards and monitoring
- [PII Guard](https://docs.xhipai.com/features/pii-guard.md): PII anonymization middleware for agent messages
- [Agent Handoff](https://docs.xhipai.com/handoff/overview.md): Transfer conversations between specialist agents mid-conversation
- [Introduction](https://docs.xhipai.com/introduction.md): RadarOS is a TypeScript-native agent orchestration framework for Node.js. Model-agnostic, multi-agent teams, workflows, RAG, and more.
- [Embeddings](https://docs.xhipai.com/knowledge/embeddings.md): Generate vector embeddings with OpenAI and Google providers
- [Hybrid Search](https://docs.xhipai.com/knowledge/hybrid-search.md): Combine vector (semantic) and keyword (BM25) search with Reciprocal Rank Fusion for better retrieval accuracy.
- [Knowledge Base Overview](https://docs.xhipai.com/knowledge/overview.md): RAG concept, KnowledgeBase class, and asTool() for equipping agents with document retrieval in RadarOS.
- [RAG Example](https://docs.xhipai.com/knowledge/rag-example.md): End-to-end Retrieval Augmented Generation walkthrough
- [Vector Stores](https://docs.xhipai.com/knowledge/vector-stores.md): VectorStore interface and implementations—InMemory, PgVector, Qdrant, MongoDB—for semantic search in RadarOS.
- [MCP Admin API](https://docs.xhipai.com/mcp/admin.md): Manage MCP server connections at runtime via REST endpoints — add, connect, disconnect, and remove MCP servers from a UI or API client.
- [MCP Client](https://docs.xhipai.com/mcp/overview.md): Connect to MCP (Model Context Protocol) servers and use their tools as native RadarOS agent tools.
- [Context Budget](https://docs.xhipai.com/memory/budgeting.md): Token-aware budget allocation for memory context injection.
- [Memory Curator](https://docs.xhipai.com/memory/curator.md): Maintenance operations for memory stores — prune, deduplicate, consolidate, and clear.
- [Graph Memory](https://docs.xhipai.com/memory/graph-memory.md): Knowledge graph with entity-relationship tracking and temporal awareness.
- [Memory Overview](https://docs.xhipai.com/memory/overview.md): Unified memory system for persistent, cross-session agent intelligence.
- [Procedural Memory](https://docs.xhipai.com/memory/procedures.md): Record and reuse successful multi-step workflows automatically.
- [Composite Scoring](https://docs.xhipai.com/memory/scoring.md): Rank memories by blending semantic relevance, recency, and importance.
- [Cross-Agent Memory Sharing](https://docs.xhipai.com/memory/sharing.md): Share memory across team members for coordinated multi-agent intelligence.
- [Simplified API](https://docs.xhipai.com/memory/simplified-api.md): Three convenience methods for quick memory operations — remember, recall, forget.
- [Memory Stores](https://docs.xhipai.com/memory/stores.md): Deep dive into each memory subsystem — summaries, user facts, user profile, entities, decisions, and learnings.
- [Temporal Awareness](https://docs.xhipai.com/memory/temporal.md): How memory tracks fact validity over time and handles contradictions.
- [Anthropic](https://docs.xhipai.com/models/anthropic.md): Use Anthropic Claude models (Claude Sonnet, Claude Haiku) with RadarOS. Setup, factory, and config.
- [AWS Bedrock](https://docs.xhipai.com/models/aws-bedrock.md): Use AWS Bedrock foundation models (Mistral, Amazon Nova, Llama, Cohere) with RadarOS. Setup, factory, config, and multi-model support.
- [AWS Claude](https://docs.xhipai.com/models/aws-claude.md): Use Claude models through AWS Bedrock with RadarOS. Native Anthropic SDK integration with AWS authentication.
- [Azure AI Foundry](https://docs.xhipai.com/models/azure-foundry.md): Use open-source models (Phi, Llama, Mistral, Cohere) hosted on Azure AI Foundry with RadarOS.
- [Azure OpenAI](https://docs.xhipai.com/models/azure-openai.md): Use OpenAI models (GPT-4o, o-series) through Azure with RadarOS. Enterprise-grade with Azure AD authentication.
- [Cohere](https://docs.xhipai.com/models/cohere.md): Use Cohere Command models with RadarOS — RAG-optimized, tool calling, and fine-tuning support.
- [Custom Provider](https://docs.xhipai.com/models/custom-provider.md): Implement your own ModelProvider for any LLM API. Register with ModelRegistry and use with agents, teams, and workflows.
- [DeepSeek](https://docs.xhipai.com/models/deepseek.md): Use DeepSeek reasoning and chat models with RadarOS agents.
- [Model Fallback & Circuit Breaker](https://docs.xhipai.com/models/fallback.md): Automatic provider failover with circuit breaker pattern for production LLM resilience
- [Google Gemini](https://docs.xhipai.com/models/google.md): Use Google Gemini models (Gemini 2.5 Flash, Gemini 2.5 Pro) with RadarOS. Setup, factory, config, and multi-modal support.
- [Meta (Llama)](https://docs.xhipai.com/models/meta.md): Use Meta Llama models with RadarOS — powerful open models via the Llama API.
- [Mistral](https://docs.xhipai.com/models/mistral.md): Use Mistral AI models with RadarOS — code generation, reasoning, and vision.
- [Ollama (Local Models)](https://docs.xhipai.com/models/ollama.md): Run local LLMs with Ollama and RadarOS. No API key needed. Llama, CodeLlama, Mistral, and more.
- [OpenAI](https://docs.xhipai.com/models/openai.md): Use OpenAI models (GPT-4o, GPT-4o-mini, o1) with RadarOS. Setup, factory, config, and per-request API key override.
- [OpenAI-Compatible](https://docs.xhipai.com/models/openai-like.md): Connect to any OpenAI-compatible API endpoint — Together, Groq, Fireworks, OpenRouter, NVIDIA, LM Studio, and more.
- [Models](https://docs.xhipai.com/models/overview.md): Model-agnostic design, ModelProvider interface, factory functions, and ModelConfig options in RadarOS.
- [Perplexity](https://docs.xhipai.com/models/perplexity.md): Use Perplexity models with built-in web search, citations, and domain filtering in RadarOS agents.
- [Smart Model Router](https://docs.xhipai.com/models/router.md): Automatically route requests to the cheapest capable model based on complexity
- [Models Type Reference](https://docs.xhipai.com/models/types-reference.md): Complete reference for ModelProvider, ModelConfig, factory function configs, and all provider-specific options.
- [Vercel v0](https://docs.xhipai.com/models/vercel.md): Use Vercel v0 models with RadarOS — optimized for web development and code generation.
- [Vertex AI](https://docs.xhipai.com/models/vertex.md): Use Google Vertex AI with RadarOS. Enterprise-grade Gemini models with Google Cloud authentication.
- [xAI (Grok)](https://docs.xhipai.com/models/xai.md): Use xAI Grok models with RadarOS agents — fast, capable, with live web search.
- [Observability](https://docs.xhipai.com/observability/overview.md): Tracing, metrics, and structured logging for RadarOS agents
- [Performance](https://docs.xhipai.com/performance.md): RadarOS performance benchmarks, optimizations, and comparison with LangChain and Agno.
- [Queue Overview](https://docs.xhipai.com/queue/overview.md): Background jobs for agents and workflows. BullMQ and Redis architecture for async processing in RadarOS.
- [Producer](https://docs.xhipai.com/queue/producer.md): Enqueue agent and workflow jobs for background processing
- [Team Scheduling](https://docs.xhipai.com/queue/teams.md): Schedule and queue team runs alongside agents and workflows
- [Worker](https://docs.xhipai.com/queue/worker.md): Process background agent and workflow jobs
- [Quickstart](https://docs.xhipai.com/quickstart.md): Get up and running with RadarOS in minutes. Install, create your first agent, add tools, and stream responses.
- [Scheduled Agents](https://docs.xhipai.com/schedules/overview.md): Run agents and workflows on a cron schedule — daily reports, periodic syncs, health checks.
- [First-Class Scheduling](https://docs.xhipai.com/scheduling/overview.md): Cron-based agent scheduling and event-driven triggers with context continuity
- [Security](https://docs.xhipai.com/security.md): Security hardening, best practices, and built-in protections across RadarOS.
- [Learned Skills](https://docs.xhipai.com/skills/learned-skills.md): Agent-learned multi-step workflows saved for future replay.
- [Skills Overview](https://docs.xhipai.com/skills/overview.md): Pre-packaged tool bundles and learned behaviors for agents.
- [DynamoDB Storage](https://docs.xhipai.com/storage/dynamodb.md): AWS DynamoDB storage driver with on-demand billing
- [In-Memory Storage](https://docs.xhipai.com/storage/in-memory.md): InMemoryStorage driver—zero dependencies, no persistence. Default for agents when no storage is configured.
- [MongoDB Storage](https://docs.xhipai.com/storage/mongodb.md): MongoDBStorage driver for document-based persistence. Requires mongodb. Call initialize() before use.
- [MySQL Storage](https://docs.xhipai.com/storage/mysql.md): MySQL-backed storage driver for sessions, memory, and culture
- [Storage Overview](https://docs.xhipai.com/storage/overview.md): StorageDriver interface, choosing a driver, and comparison of InMemory, SQLite, PostgreSQL, and MongoDB storage backends in RadarOS.
- [PostgreSQL Storage](https://docs.xhipai.com/storage/postgres.md): PostgresStorage driver for production persistence. Requires pg. Call initialize() before use.
- [Redis Storage](https://docs.xhipai.com/storage/redis.md): Redis-backed storage driver with optional TTL
- [SQLite Storage](https://docs.xhipai.com/storage/sqlite.md): SqliteStorage driver for file-based persistence. Requires better-sqlite3. No initialize() needed.
- [System Architecture](https://docs.xhipai.com/system-architecture.md): RadarOS architecture overview.
- [Broadcast Mode](https://docs.xhipai.com/teams/broadcast.md): All agents process input; results are merged. Example with multi-perspective analysis.
- [Collaborate Mode](https://docs.xhipai.com/teams/collaborate.md): Iterative rounds between agents. maxRounds config. Example with iterative refinement.
- [Coordinate Mode](https://docs.xhipai.com/teams/coordinate.md): Orchestrator picks which member agent handles each task. Example with researcher, writer, and reviewer.
- [Teams](https://docs.xhipai.com/teams/overview.md): Multi-agent collaboration in RadarOS. TeamMode, TeamConfig, coordinate, route, broadcast, and collaborate.
- [Route Mode](https://docs.xhipai.com/teams/route.md): Route to the single best agent based on input. Example with specialist agents.
- [Calculator](https://docs.xhipai.com/toolkits/calculator.md): Evaluate math expressions safely — no API key required.
- [Google Calendar](https://docs.xhipai.com/toolkits/calendar.md): List, create, get, and delete Google Calendar events.
- [Code Interpreter](https://docs.xhipai.com/toolkits/code-interpreter.md): Execute JavaScript, Python, or TypeScript code in a sandboxed subprocess.
- [Discord](https://docs.xhipai.com/toolkits/discord.md): Send messages, read messages, and list channels in Discord servers.
- [DuckDuckGo](https://docs.xhipai.com/toolkits/duckduckgo.md): Search the web and get news using DuckDuckGo — no API key required.
- [File System](https://docs.xhipai.com/toolkits/filesystem.md): Read, write, list, and inspect local files with path sandboxing.
- [Git](https://docs.xhipai.com/toolkits/git.md): Local git operations — status, diff, log, commit, branch.
- [GitHub](https://docs.xhipai.com/toolkits/github.md): Interact with GitHub repos, issues, PRs, and file content.
- [Gmail](https://docs.xhipai.com/toolkits/gmail.md): Send, search, and read emails via the Gmail API.
- [Google Sheets](https://docs.xhipai.com/toolkits/google-sheets.md): Read, write, and append data in Google Sheets.
- [Google Workspace](https://docs.xhipai.com/toolkits/google-workspace.md): Access 30+ Google Workspace APIs (Drive, Gmail, Calendar, Sheets, Docs, Chat, Admin, and more) via the gws CLI.
- [Hacker News](https://docs.xhipai.com/toolkits/hackernews.md): Fetch top stories and user details from Hacker News — no API key required.
- [HTTP / REST](https://docs.xhipai.com/toolkits/http.md): Make arbitrary HTTP requests from your agent — call any API.
- [Image Generation](https://docs.xhipai.com/toolkits/image-generation.md): Generate and edit images via OpenAI DALL-E API.
- [Jira](https://docs.xhipai.com/toolkits/jira.md): Search, create, and update Jira issues from your agent.
- [Notion](https://docs.xhipai.com/toolkits/notion.md): Search, read, and create pages in Notion databases.
- [Toolkits](https://docs.xhipai.com/toolkits/overview.md): Pre-built tool collections for common integrations — web search, email, messaging, databases, and more.
- [PageIndex](https://docs.xhipai.com/toolkits/pageindex.md): Vectorless, reasoning-based RAG — submit PDFs, build tree indexes, and retrieve via LLM reasoning.
- [PDF](https://docs.xhipai.com/toolkits/pdf.md): Extract text, metadata, and page content from PDF files.
- [Redis](https://docs.xhipai.com/toolkits/redis.md): Get, set, delete, list, and increment keys in Redis.
- [S3 Cloud Storage](https://docs.xhipai.com/toolkits/s3.md): Upload, download, list, delete, and presign URLs for S3-compatible storage.
- [Web Scraper](https://docs.xhipai.com/toolkits/scraper.md): Extract text content and links from web pages — no dependencies.
- [Shell](https://docs.xhipai.com/toolkits/shell.md): Execute shell commands with timeout, output limits, and command allowlisting.
- [Slack](https://docs.xhipai.com/toolkits/slack.md): Send and read messages, list channels, and reply in threads.
- [SQL Database](https://docs.xhipai.com/toolkits/sql.md): Query databases (SQLite, PostgreSQL, MySQL) from your agent.
- [Stripe](https://docs.xhipai.com/toolkits/stripe.md): List charges, get customers, create refunds, list subscriptions, and get invoices.
- [Telegram](https://docs.xhipai.com/toolkits/telegram.md): Send messages, photos, and read updates via the Telegram Bot API.
- [Web Search](https://docs.xhipai.com/toolkits/websearch.md): AI-optimized web search using Tavily or SerpAPI backends.
- [WhatsApp](https://docs.xhipai.com/toolkits/whatsapp.md): Send messages via Meta's WhatsApp Business Cloud API.
- [Wikipedia](https://docs.xhipai.com/toolkits/wikipedia.md): Search and read Wikipedia articles — no API key required.
- [YouTube](https://docs.xhipai.com/toolkits/youtube.md): Search YouTube videos and extract transcripts/captions.
- [API Keys](https://docs.xhipai.com/transport/api-keys.md): Per-request API keys via headers and request body for OpenAI, Google, and Anthropic in RadarOS transport.
- [Express](https://docs.xhipai.com/transport/express.md): Expose agents, teams, and workflows as REST API endpoints
- [File Upload](https://docs.xhipai.com/transport/file-upload.md): File upload middleware and buildMultiModalInput for sending images, audio, and documents to agents via HTTP.
- [JWT & RBAC Middleware](https://docs.xhipai.com/transport/jwt-rbac.md): Plug-and-play authentication and role-based access control
- [Transport Overview](https://docs.xhipai.com/transport/overview.md): Transport layer concept—Express REST API and Socket.IO for exposing RadarOS agents, teams, and workflows over HTTP and WebSockets.
- [Socket.IO Gateway](https://docs.xhipai.com/transport/socketio.md): createAgentGateway for real-time agent streaming. Events, namespace, auth, and a chat example.
- [Swagger UI](https://docs.xhipai.com/transport/swagger.md): Swagger UI setup, OpenAPI spec generation, and SwaggerOptions for documenting RadarOS agent APIs.
- [Voice Agents](https://docs.xhipai.com/voice/overview.md): Build real-time voice agents with speech-to-speech AI. Supports OpenAI Realtime and Google Gemini Live APIs.
- [Vision Agents](https://docs.xhipai.com/voice/vision-agents.md): Build real-time audio + video agents that can see, hear, and speak — with voice selection, multilingual auto-detection, screen sharing, camera feeds, and interruption support.
- [Webhooks & Event Destinations](https://docs.xhipai.com/webhooks/overview.md): Push agent events to external systems like HTTP endpoints, Slack, and email
- [Workflows](https://docs.xhipai.com/workflows/overview.md): Stateful step-based execution in RadarOS. WorkflowConfig, state machine concept, and basic example.
- [Retry & Error Handling](https://docs.xhipai.com/workflows/retry-policy.md): retryPolicy config (maxRetries, backoffMs), StepResult status (done, error, skipped), WorkflowResult.
- [Workflow Steps](https://docs.xhipai.com/workflows/steps.md): AgentStep, FunctionStep, ConditionStep, and ParallelStep in RadarOS workflows. Code examples for each.

## Optional

- [GitHub](https://github.com/xhipment/radar-os)
- [npm](https://www.npmjs.com/org/radaros)