Claude Cookbook: Practical Guides and Community Feedback

Claude Cookbook: Practical Guides and Community Feedback

What Is the Claude Cookbook?

The Claude Cookbook is a collection of practical guides and examples for using Claude effectively, hosted at https://platform.claude.com/cookbook/. It provides step‑by‑step notebooks and code snippets that demonstrate how to apply Claude’s capabilities in areas such as agent orchestration, tool use, evaluations, Retrieval‑Augmented Generation, and multimodal tasks.

Categories and Sample Guides

The cookbook organizes its entries into thematic categories. Each category groups guides that share a common focus, making it easier to locate relevant patterns.

  • Evals Tools – Guides that show how to evaluate Claude’s performance, such as reproducing agentic search benchmark scores or generating synthetic test data for prompt templates.
  • Agent Patterns – Recipes for common multi‑agent workflows, including async multi‑agent orchestration, basic workflows, orchestrator‑workers, and evaluator‑optimizer patterns.
  • Claude Agent SDK – Tutorials that walk through building agents with the SDK, covering topics like the one‑liner research agent, site reliability agent, vulnerability detection agent, and hosting agents on Docker, Modal, or Kubernetes.
  • Claude Managed Agents Tools – Examples that extend managed agents with memory, user preference storage, data analysis, Slack integration, and SRE incident response.
  • Claude Managed Agents Evals – Guides that combine a drafts feedback loops.
  • Tools Agent Patterns – Strategies for context engineering, automatic context compaction, programmatic tool calling, and tool search with embeddings.
  • RAG & Retrieval – Techniques for building knowledge graphs, text‑to‑SQL conversion, contextual embeddings, summarization, and classification with Claude.
  • Integrations – Guides that connect Claude to external services such as ElevenLabs for low‑latency voice, Deepgram for audio transcription, Wolfram Alpha for computational queries, and LlamaIndex for multimodal RAG pipelines.
  • Multimodal Tools – Recipes that add vision capabilities, such as a crop tool for image analysis, best practices for vision, and combining vision with tools to extract structured data from images.
  • Responses – Prompting guides for frontend aesthetics, JSON mode, speculative prompt caching, batch processing, and sampling beyond max tokens.
  • Skills – Notebooks that teach how to use Claude’s built‑in Excel, PowerPoint, and PDF skills, and how to create custom skills for specialized workflows.
  • Thinking – Guides on extended thinking, both alone and combined with tool use, for transparent step‑by‑step reasoning.

Each entry lists its title, categories, author(s), and month/year of publication (e.g., “Reproduce Claude's agentic search benchmark scores in the Messages API – Jun 2026” or “Hosting your agent – May 2026”).

Notable Guides by Category

Below are a few representative cookbooks that illustrate the depth of the collection.

  • Agentic Search Benchmark (Evals Tools, Jun 2026) – Build a Messages API harness that reproduces published DeepSearchQA and BrowseComp scores using programmatic tool calling, server‑side compaction, and task budgets.
  • Classifier fallback and billing for Claude Fable 5 (Responses Safeguards Billing, Jun 2026) – Detect safety classifier blocks on Fable 5 and fall back to Opus 4.8 with server‑side or SDK‑based client‑side fallback, including streaming behavior and new billing changes.
  • Async multi‑agent orchestration (Agent Patterns, Jun 2026) – Two patterns: a fixed N‑agent team with peer messaging through a shared hub, and dynamically spawned async subagents, reduced to bare messaging and lifecycle mechanics.
  • Hosting your agent (Claude Agent SDK, May 2026) – Deploy the research agent from notebook 00 through three tiers of operational maturity (Docker, Modal, Kubernetes) with the same container image and HTTP interface at every tier.
  • Multiagent: coordinate a specialist team (Claude Managed Agents Tools, May 2026) – Use the multiagent coordinator config to run three specialists (web‑search researcher, file‑reading librarian, rules‑based pricer) with scoped toolsets to assemble a sales proposal.
  • Outcomes: agents that verify their own work (Claude Managed Agents Evals, May 2026) – Build a grade‑and‑revise loop where a writer drafts a cited research brief, a stateless grader checks every quote against a rubric, and feedback drives revisions until the brief passes.
  • Build agents that remember your users (Claude Managed Agents Tools, Apr 2026) – Give Claude Managed Agents a Memory store so they learn and remember user preferences across multiple interactions.
  • The vulnerability detection agent (Claude Agent SDK Cybersecurity, Apr 2026) – Build a vulnerability‑discovery agent that threat‑models a C target, hunts memory‑safety bugs with built‑in file tools, and triages findings into a structured report.
  • Build an SRE incident response agent (Claude Managed Agents Observability, Apr 2026) – Wire Claude into on‑call flow: when an alert fires, the agent reads logs and runbooks, pinpoints root cause, opens a fix PR, and waits for approval before merging.
  • Build a data analyst agent (Claude Managed Agents Tools, Apr 2026) – Turn a CSV into a narrative HTML report with interactive charts using a sandboxed environment and file mounting.
  • Build a Slack data analyst bot (Claude Managed Agents Integrations, Apr 2026) – Mention the bot with a CSV to get an analysis report in‑thread, with multi‑turn follow‑ups on the same session.
  • Managed Agents tutorial: iterate on a failing test suite (Claude Managed Agents Tools, Apr 2026) – Entry‑point tutorial that walks through agent/environment/session creation, file mounts, and the streaming event loop by fixing three planted bugs in a calc.py package.
  • Managed Agents tutorial: production setup (Claude Managed Agents Integrations, Apr 2026) – End‑to‑end production story covering vault‑backed MCP credentials, the session.status_idled webhook pattern for human‑in‑the‑loop, and resource lifecycle CRUD verbs.
  • Managed Agents tutorial: prompt versioning and rollback (Claude Managed Agents Evals, Apr 2026) – Server‑side prompt versioning: create v1, evaluate against a labelled test set, ship v2, detect regression, roll back by pinning sessions to version 1.
  • Threat intelligence enrichment agent (Tools Agent Patterns Cybersecurity, Apr 2026) – Autonomous agent that investigates IOCs by querying multiple threat intel sources, cross‑referencing findings, mapping to MITRE ATT&CK, and producing structured reports for SIEM and SOAR integration.
  • Building a session browser (Claude Agent SDK Agent Patterns, Mar 2026) – List, read, rename, tag, and fork Agent SDK sessions on disk to build a conversation history sidebar without writing a transcript parser.
  • Knowledge graph construction with Claude (RAG & Retrieval Tools, Mar 2026) – Build knowledge graphs from unstructured text using Claude for entity extraction, relation mining, deduplication, and multi‑hop graph querying.
  • Context engineering: memory, compaction, and tool clearing (Tools Agent Patterns, Mar 2026) – Compare context engineering strategies for long‑running agents, learn when each applies, what it costs, and how they compose.
  • Migrating from the OpenAI Agents SDK (Claude Agent SDK Agent Patterns, Mar 2026) – Port an OpenAI Agents SDK app to the Claude Agent SDK, mapping each primitive (tools, guardrails, sessions, handoffs) through a single expense‑approval agent example.
  • The site reliability agent (Claude Agent SDK Agent Patterns, Feb 2026) – Build an incident response agent with read‑write MCP tools for autonomous diagnosis, remediation, and post‑mortem documentation.
  • Session memory compaction (Agent Patterns Responses, Jan 2026) – Manage long‑running Claude conversations with instant session memory compaction using background threading and prompt caching.
  • Programmatic tool calling (PTC) (Tools, Nov 2025) – Reduce latency and token consumption by letting Claude write code that calls tools programmatically in the code execution environment.
  • Tool search with embeddings (Tools RAG & Retrieval, Nov 2025) – Scale Claude applications to thousands of tools using semantic embeddings for dynamic tool discovery.
  • Automatic context compaction (Tools Agent Patterns, Nov 2025) – Manage context limits in long‑running agentic workflows by automatically compressing conversation history.
  • Low latency voice assistant with ElevenLabs (Integrations, Nov 2025) – Build a low‑latency voice assistant using ElevenLabs for speech‑to‑text and text‑to‑speech combined with Claude.
  • Giving Claude a crop tool for better image analysis (Multimodal Tools, Nov 2025) – Give Claude a crop tool to zoom into image regions for detailed analysis of charts, documents, and diagrams.
  • Prompting for frontend aesthetics (Responses Skills, Oct 2025) – Guide to prompting Claude for distinctive, polished frontend designs avoiding generic aesthetics.
  • Claude Skills for financial applications (Skills, Oct 2025) – Build financial dashboards and portfolio analytics using Claude’s Excel, PowerPoint, PDF skills.
  • Building custom Skills for Claude (Skills, Oct 2025) – Create, deploy, and manage custom skills extending Claude with specialized organizational workflows.
  • Introduction to Claude Skills (Skills, Oct 2025) – Create documents, analyze data, automate workflows with Claude’s Excel, PowerPoint, PDF skills.
  • The one‑liner research agent (Claude Agent SDK Agent Patterns, Sep 2025) – Build a research agent using Claude Code SDK with WebSearch for autonomous research.
  • The chief of staff agent (Claude Agent SDK Agent Patterns, Sep 2025) – Build multi‑agent systems with subagents, hooks, output styles, and plan mode features.
  • The observability agent (Claude Agent SDK Agent Patterns, Sep 2025) – Connect agents to external systems via MCP servers for GitHub monitoring and CI workflows.
  • Tool evaluation (Evals, Sep 2025) – Run parallel agent evaluations on tools independently from evaluation task files.
  • Usage & cost Admin API cookbook (Observability, Aug 2025) – Programmatically access and analyze Claude API usage and cost data via Admin API.
  • Memory & context management with Claude Sonnet 4.6 (Tools Agent Patterns, May 2025) – Build AI agents with persistent memory using Claude’s memory tool and context editing.
  • Speculative prompt caching (Responses, May 2025) – Reduce time‑to‑first‑token by warming cache speculatively while users formulate their queries.
  • Parallel tool calls on Claude 3.7 Sonnet (Tools, Mar 2025) – Enable parallel tool calls on Claude 3.7 Sonnet using batch tool meta‑pattern workaround.
  • Extended thinking (Thinking, Feb 2025) – Use Claude’s extended thinking for transparent step‑by‑step reasoning with budget management.
  • Extended thinking with tool use (Thinking Tools, Feb 2025) – Combine extended thinking with tools for transparent reasoning during multi‑step workflows.
  • Basic workflows (Agent Patterns, Dec 2024) – Three simple multi‑LLM workflow patterns trading cost or latency for improved performance.
  • Evaluator optimizer (Agent Patterns Evals, Dec 2024) – Workflow pattern using one LLM for generation and another for evaluation feedback loop.
  • Orchestrator workers (Agent Patterns, Dec 2024) – Central LLM dynamically delegates tasks to worker LLMs and synthesizes their combined results.
  • Batch processing with Message Batches API (Responses, Oct 2024) – Process large volumes of Claude requests asynchronously with 50% cost reduction using batches.
  • Text to SQL with Claude (RAG & Retrieval, Sep 2024) – Convert natural language queries to SQL using RAG, chain‑of‑thought, and self‑improvement techniques.
  • Enhancing RAG with contextual retrieval (RAG & Retrieval, Sep 2024) – Improve RAG accuracy by adding context to chunks before embedding with prompt caching.
  • Summarization with Claude (RAG & Retrieval Responses, Aug 2024) – Comprehensive guide to summarizing legal documents with evaluation and advanced techniques.
  • Retrieval augmented generation (RAG & Retrieval, Jul 2024) – Build and optimize RAG systems with Claude using summary indexing and reranking techniques.
  • Classification with Claude (RAG & Retrieval, May 2024) – Build classification systems with Claude using RAG and chain‑of‑thought for insurance tickets.
  • Tool choice (Tools, May 2024) – Control how Claude selects tools using tool_choice parameter for forced or auto selection.
  • Using vision with tools (Multimodal Tools, May 2024) – Combine Claude's vision with tools to extract structured data from images like nutrition labels.
  • Sampling responses from Claude beyond the max tokens limit (Responses, May 2024) – Generate longer responses beyond max_tokens limit using prefill technique with message continuation.
  • Best practices for using vision with Claude (Multimodal, May 2024) – Tips and techniques for optimal image processing performance with Claude's vision capabilities.
  • Note‑saving tool with Pydantic and Anthropic tool use (Tools, Apr 2024) – Create validated tools using Pydantic models for type‑safe Claude tool use interactions.
  • Transcribe an audio file with Deepgram & use Anthropic to prepare interview questions! (Integrations Multimodal, Apr 2024) – Transcribe audio with Deepgram and generate interview questions using Claude for preparation.
  • Using the Wolfram Alpha LLM API as a tool with Claude (Integrations Tools, Apr 2024) – Integrate Wolfram Alpha LLM API as Claude tool for computational queries and answers.
  • Using a calculator tool with Claude (Tools, Apr 2024) – Provide Claude with calculator tool for arithmetic operations and mathematical problem solving.
  • Creating a customer service agent with client‑side tools (Tools Agent Patterns, Apr 2024) – Build customer service chatbot with Claude using tools for customer lookup and order management.
  • Extracting structured JSON using Claude and tool use (Responses Tools, Apr 2024) – Extract structured JSON data from various inputs using Claude's tool use capabilities.
  • Metaprompt (Responses, Mar 2024) – Prompt engineering tool that generates starting prompts for your tasks to solve blank‑page problem.
  • Citations (Responses RAG & Retrieval, Mar 2024) – Enable Claude to provide detailed source citations when answering document‑based questions for verification.
  • Claude 3 RAG agents with LangChain v1 (Integrations RAG & Retrieval Agent Patterns, Mar 2024) – Build RAG agents with Claude 3 using LangChain v1's updated agent framework patterns.
  • Summarizing web page content with Claude 3 Haiku (RAG & Retrieval, Mar 2024) – Fetch and summarize web page content using Claude 3 Haiku via URL extraction.
  • Using Haiku as a sub‑agent (Agent Patterns, Mar 2024) – Analyze financial reports using Haiku sub‑agents for extraction and Opus for synthesis.
  • Multi‑modal (Integrations Multimodal, Mar 2024) – Use LlamaIndex's Anthropic MultiModal LLM abstraction for image understanding and reasoning.
  • How to build a RAG system using Claude 3 and MongoDB (Integrations RAG & Retrieval, Mar 2024) – Build chatbot RAG system with Claude and MongoDB using tech news as knowledge base.
  • Building evals (Evals, Mar 2024) – Build robust evaluation systems to measure and improve Claude's performance on key metrics.
  • Building a moderation filter with Claude (Responses, Mar 2024) – Build customizable content moderation filters by defining rules and categories in prompts.
  • Prompting Claude for "JSON mode" (Responses, Mar 2024) – Get reliable JSON output from Claude using effective prompting techniques without constrained sampling.
  • How to make SQL queries with Claude (RAG & Retrieval, Mar 2024) – Generate SQL queries from natural language questions using Claude with database schema context.
  • Getting started - how to pass images into Claude (Multimodal, Mar 2024) – Tutorial on passing images to Claude 3 API for vision‑based text analysis.
  • How to transcribe documents with Claude (Multimodal, Mar 2024) – Extract and structure unstructured text from images and PDFs using Claude 3's vision.
  • Working with charts, graphs, and slide decks (Multimodal, Mar 2024) – Extract insights from charts, graphs, and presentations using Claude's vision analysis capabilities.
  • Multi‑document agents (Integrations RAG & Retrieval Agent Patterns, Mar 2024) – Build RAG for large document collections using DocumentAgents with ReAct Agent pattern.
  • ReAct agent (Integrations Agent Patterns Tools, Mar 2024) – Create ReAct agents with LlamaIndex for tool‑based reasoning and action workflows.
  • RAG pipeline with LlamaIndex (Integrations RAG & Retrieval, Mar 2024) – Build basic RAG pipeline with LlamaIndex for document retrieval and question answering.
  • RouterQuery engine (Integrations RAG & Retrieval, Mar 2024) – Route queries to different indices using LlamaIndex RouterQueryEngine for multi‑document search.
  • SubQuestionQueryEngine (Integrations RAG & Retrieval, Mar 2024) – Decompose complex queries into sub‑questions across multiple documents using LlamaIndex engine.
  • Retrieval‑augmented generation using Pinecone (Integrations RAG & Retrieval, Feb 2024) – Connect Claude with Pinecone vector database for retrieval‑augmented generation and semantic search.
  • "Uploading" PDFs to Claude via the API (RAG & Retrieval, Aug 2023) – Process and summarize PDF documents using Claude API with text extraction and encoding.
  • Iteratively searching Wikipedia with Claude (Integrations, Aug 2023) – Legacy notebook showing iterative Wikipedia searches with Claude 2 for research workflows.

Community Reception on Hacker News

The Hacker News post linking to the Claude Cookbook received 330 points and 164 comments. Reactions varied widely, reflecting both enthusiasm for specific technical recipes and skepticism about the cookbook’s overall utility.

Some commenters appreciated particular sections. For example, one user praised Matt Pocock’s skills, noting they are designed to be called by the user rather than automatically, which reduces context usage and encourages thinking about the end result:

"I've been using Matt Pocock's skills at home, and they seem pretty great. He makes a great case for them vs other skills in his videos. I don't remember the exact arguments, but basically they're designed to be called by the user instead of called automatically, and that makes them take less context merely because they exist. I've also found that they tend to coerce the programmer into thinking about the end result, rather than try to push them out of that role." – @wccrawford

Another user highlighted the “prompting for frontend aesthetics” guide, though they questioned its effectiveness:

"The before and after images in “prompting for frontend aesthetics” are hilarious. Did no one look at these to make sure the skill actually improved the design?" – @semiquaver "Before: bland After: bland with gradients" – @gherkinnn

Several commenters compared the Claude Cookbook to existing resources from other labs:

"Thanks! I also like the OpenAI Cookbook: https://developers.openai.com/cookbook Other AI labs also tend to publish examples and cookbooks on GitHub and Hugging Face, so it's always worth keeping an eye on those as well." – @beklein

A few users misunderstood the nature of the cookbook, expecting literal cooking recipes:

"I thought this is a new product to generate recipes with Claude." – @brkn "I thought it would be a litteral cookbook on how to make plausible/feasible recipes with LLMs (we're not here yet sadly)." – @AntiqueFig "Immensely disappointed; I expected an actual cookbook." – @lkm0 "Based on the title, I was hoping this was going to be a list of popular recipes Claude has suggested for cooking food." – @mbreese

Some commenters expressed broader criticisms about the relevance of agent‑focused guides:

"Gotta be honest, almost every “how to use AI” resource seems pointless to me. I'm either going to ask the AI how to do it, or if it's about using the AI then we can just bake it into the harness or wait for Anthropic/OpenAI to do it for me because they're always trivial. All of these resources on agentic workflows, managing agent memory, harness engineering, etc. appear to just be theatre to me." – @mindwok

Others raised concerns about curation and quality, suggesting the collection needs more rigorous editorial oversight:

"Some of these are pretty good, especially for shopping for solutions I'd never considered. But in general there is a quality problem: most don't bother trying to justify their inclusion in the recipe list much less explain what compelled the recipe. These really need a human preamble on why they exist. Like, is this a recipe you've been using for months in production, or is it just what Claude spits out if you ask it “how can I compact context in my chat bot?” The former is very interesting (it's what curated skill repos are), but the latter is just slop. e.g. a recipe from @Anthropic shows how to upload a pdf to the API: https://platform.claude.com/cookbook/misc-pdf-upload-summari... -- That said, it was created in 2023 back when my use of AI was still autocomplete in Cursor. Kinda needs to be an eternally pruned, curated, living collection, not an append‑only one from random people via PR." – @hombre_fatal

A few users noted UI issues:

"They could have done at least one pass on the UI. Can't even get simple table spacing right" – @merrvk

Some observed that patterns from cookbooks often get incorporated into future model releases:

"Recipes like this get baked into the next iteration of a model." – @winterbourne

Finally, a few commenters advocated learning by doing rather than relying on external guides:

"The best way to learn how to use LLM tools and harnesses is to use them." – @Arubis

Overall, the discussion shows that while the Claude Cookbook provides a rich set of technical examples—particularly around agent orchestration, tool use, and integrations—its value is judged differently depending on readers’ expectations, their preference for curated, production‑tested patterns versus exploratory examples, and their interest in non‑cooking‑related AI workflows.

Sources