Advanced | Agno

Example	Description
01 Create Cultural Knowledge	Migration notice for removed cultural-knowledge APIs.
04 Manually Add Culture	Migration notice for removed manual-culture APIs.
Advanced Compression	Set a context token based limit for tool call compression.
Agent Serialization	Serialize and deserialize agents using to_dict/from_dict and save/load with a database.
Background Execution	Background execution allows you to start an agent run that returns immediately.
Background Execution Structured	Combines background execution (non-blocking, async) with Pydantic output_schema.
Basic Agent Events	Stream agent events including run lifecycle, tool calls, and content output.
Cache Model Response	Example showing how to cache model responses to avoid redundant API calls.
Cancel Run	Example demonstrating how to cancel a running agent execution.
Compression Events	Test script to verify compression events are working correctly.
Concurrent Execution	Concurrent Agent Execution with asyncio.gather.
Custom Cancellation Manager	Shows how to extend BaseRunCancellationManager to implement your own.
Custom Logging	Example showing how to use a custom logger with Agno.
Debug	You can set the debug mode on the agent for all runs to have more verbose output.
Multi-Model Metrics	Track per-model token usage with memory model breakdown in metrics.details.
Culture Metrics	Migration notice for removed Culture metrics; links to current agent metrics.
Session Metrics	Accumulate metrics across multiple runs within a session using SessionMetrics.
Session Summary Metrics	Track session summary model token usage under the session_summary_model detail key.
Streaming Metrics	Capture metrics from streaming responses using yield_run_output=True.
Tool Call Metrics	Track tool execution timing with ToolCallMetrics on each ToolExecution.
Background Execution Metrics	Track metrics for background (async) agent runs with full token and model details.
Reasoning Agent Events	Capture reasoning lifecycle and ReasoningContentDelta events from a separate reasoning stage.
Retries	Example demonstrating how to set up retries with an Agent.
Tool Call Compression	Compress tool call history to reduce context size.
Learning Machine	Create agents that learn and improve from interactions over time.
Memory Manager	Use a MemoryManager to give agents persistent memory across sessions.
Basic Reasoning	Add chain-of-thought reasoning capabilities to agents.
Reasoning With Model	Use a separate reasoning model with configurable step limits.
Basic Skills	Load skills from a local directory with LocalSkills and use them in a code review agent.
Check Style	Check Python code for style issues.
Commit Message	Validate or generate conventional commit messages.
SSE Reconnection	Reconnect to a background agent stream after disconnection using the /resume endpoint.
Cancel Run Persistence	Cancel an agent run mid-stream and verify that partial content and messages are preserved in the database.
Combined Metrics	Migration notice for the removed combined-metrics recipe; links to current agent metrics.
Interchange Model: All 5 Providers	Cycles through OpenAI Chat, OpenAI Responses, Claude, Gemini, and AWS Claude.
Claude Gemini	Switch a single agent between Claude and Gemini mid-conversation with shared Postgres history.
OpenAI Chat Responses	Alternate one agent between OpenAIChat and OpenAIResponses mid-session over shared Postgres history.
OpenAI Claude	Swap a single agent between gpt-5.6-luna and Claude mid-session with tool-call history stored in Postgres.
OpenAI Gemini	Move one conversation between gpt-5.6-luna and Gemini across turns with history persisted in Postgres.
Metrics	Inspect per-message, run, and session metrics after a YFinance tool call on a Postgres-backed agent.
Background Execution Concurrency	Example demonstrating the concurrency limit for background runs.
Background Streaming Resume	Example demonstrating background streaming with disconnect and resume.
Redis Event Stream Resume	Example demonstrating cross-process streaming resume with RedisEventStream.