Claude Ustalığı

What is Claude? The New Generation of AI Assistants

Learn from scratch what Claude is, how it differs from traditional chatbots, and how it will reshape your daily workflow.

Anthropic, Constitutional AI and Safety Philosophy

Understand Constitutional AI, Anthropic's safety philosophy, and the reasoning behind what Claude refuses to do.

Meet the Models: Opus, Sonnet, Haiku Compared

Learn the strengths, costs, and speed of Claude's three main models (Opus, Sonnet, Haiku) and which one to pick in each scenario.

Your First Conversation: Web, Mobile, and Desktop Tour

Tour every corner of Claude.ai web, mobile, and desktop apps; master Projects, Artifacts, and keyboard shortcuts from day one.

Claude's Capability Map and Limits

See where Claude shines and where it struggles on a clear capability map — covering hallucination, fresh data, and complex math.

Clarity and Specificity: From Vague to Surgical Prompts

Vague prompts produce vague answers. Learn to surface the true intent under fuzzy requests and acquire surgical specificity.

Structured Prompts with XML Tags

Make your prompts parseable, maintainable, and testable using the XML tags pattern Anthropic recommends.

Designing and Versioning Prompt Templates

Production prompts should be treated like code: templates, parameters, versions, tests, and monitoring. Learn to manage them with software-grade discipline.

System Prompt and Persona Design

How do you design Claude's personality, boundaries, and behavior frame via the system prompt? Production-ready persona patterns inside.

Output Format Control: JSON, Markdown, Tables

Master three techniques to keep Claude's output parseable, consistent, and error-free: schema, prefill, and validator-loop.

Multi-Step Task Decomposition

Instead of solving complex tasks in one prompt, break them into modular subtasks for better accuracy, testability, and cost.

Prompt Debugging: Why Isn't This Working?

Prompt not working? Learn a systematic diagnostic tree, logging strategy, and model swap checklist.

Token Economy and Cost Optimization

Produce the same quality at 50-90% lower cost using token economy: prompt caching, model tiering, output constraints, batching.

Generating Functions and Classes from Scratch

Workflow for producing production-ready functions/classes from a clear spec with Claude: signature, edge cases, tests, docs.

Code Review, Refactor, and Optimization

Use Claude as a code reviewer: pattern detection, refactor suggestions, performance tuning, and security review.

Bug Hunt: From Stack Trace to Fix

Workflow for systematically finding a bug with Claude: minimum repro, hypothesis, isolation, fix, regression test.

Test Writing, TDD, and Coverage Strategies

Workflow for writing tests with Claude: unit, integration, property-based, snapshot. Coverage targets and which tests not to write.

Documentation, Docstrings, and README Generation

Generate docstrings, READMEs, ADRs, runbooks with Claude. Turn docs from a debt into a revenue-producing asset.

Claude Code: Working in the Terminal

Claude in the terminal via Claude Code CLI — project mapping, file editing, code execution, hooks, and plan mode.

Blog, Article, and Long-Form Content

From a topic seed to a 2,000-4,000-word quality long-form piece with Claude — including SEO, tone, and originality.

Emails, Briefs, and Business Correspondence

Cold sales emails, customer save, bad-news memos, briefs — proven prompt patterns for every kind of business writing.

Story, Script, and Character Development

Fundamentals of character, conflict, scene, and dialogue + ethics of creative writing with Claude. From novel design to short-film scripting.

Translation, Localization, and Style Consistency

Beyond word-for-word — cultural adaptation, brand voice, terminology memory, and QA cycle. Focus on Turkish-English.

Preserving Voice and Tone: Style Transfer

Transferring written voice to Claude — generating consistent output using your personal style, brand voice, or past writings.

Working with CSV, JSON, and Tables

Reliably process tabular data with Claude: cleaning, transforming, validating, generating control code, and verifying the result.

Statistical Interpretation and Visualization

Stories, not just numbers — interpret statistical results with Claude, with significance thresholds and visualization choices.

Long Document Summarization and Synthesis

From 200-page reports to a 1-page brief: map-reduce, anchored summarization, and faithfulness evaluation.

Comparative Analysis and Multi-Source Evaluation

Way to weigh conflicting sources together: source-reliability tagging, contradiction maps, synthesis bias checks.

Catching Hallucination and Verifying Sources

You can't eliminate hallucination but you can catch it. Systematize verification of Claude output with 6 techniques.

Tool Use: Granting Claude Real Capabilities

How to teach Claude to use a calculator, database, email, Slack, code sandbox? Anatomy of tool use and production patterns.

Vision: Image Understanding and Analysis

Screenshots, photos, charts, handwritten notes — extracting information from images with Claude vision, plus its limits.

PDF and Document Processing

Sending PDFs to Claude, extracting data from multi-page documents, and forms / contracts analysis.

Computer Use: Screen and Browser Control

Two capabilities where Claude controls your screen, browser, mouse and keyboard: Computer Use and Claude in Chrome. Includes safe-use practices.

Extended Thinking: Deep Reasoning Mode

Open the model's hidden thinking buffer on complex tasks: what extended thinking is, when to enable, and the cost.

Artifacts: Live Working Outputs

Mechanics of Artifacts — Claude's code, SVG, or React component live-rendered in a side panel. Build and iterate live.

Getting Started with the API: Auth, First Request, SDK Setup

Grab your API key from Anthropic console, install the SDK, run your first call. Python and TypeScript step by step.

Messages API: Multi-Turn Conversations

Building blocks of Messages API, system message, multi-turn conversation, response object, and token counting.

Streaming Responses and Real-Time UX

What streaming is, why it matters, and integration with SSE / Web Streams.

Tool Use API: Function Calling in Practice

Finish tool use over the API: full loop, parallel tools, error feedback, and schema validation.

Cut Cost up to 90% with Prompt Caching

Cache stable system prompts, large few-shot blocks, and long documents to slash input cost.

Batch API: Bulk Async Workloads

Run async workloads at 50% cost with the Batch API. Ideal for labeling, content generation, and evals.

Error Handling, Rate Limits, and Retry Strategies

Classify API errors, apply the right retry strategy, use idempotency keys, design DLQs.

Eval Sets and LLM-as-Judge

Design eval sets that measure production quality: building, balancing, auto-scoring (LLM-as-judge), human calibration.

Prompt Injection, Jailbreak, and Defense

How adversarial users, malicious content, or manipulated data affect Claude — and eight defense patterns.

Cost Monitoring, Quotas, and Budget Alarms

Avoid bill surprises with a cost pipeline: per-user quotas, alarm thresholds, anomaly detection.

Latency, Caching, and Performance Optimization

Eight levers to reduce p50/p95/p99: model choice, caching, streaming, parallelism.

Logging, Tracing, and Observability

OpenTelemetry-compatible LLM tracing, structured log schema, error reporting, prompt-versioned observability.

What Is an Agent? Reactive vs Autonomous

Agent vs pipeline, the four building blocks (planner, memory, tools, controller), and which problems require an agent.

Build Your First Agent with the Claude Agent SDK

Hello-World agent with the Claude Agent SDK: tool definitions, system prompt, controller loop, and human approval.

Multi-Tool Agent Architectures (Planner-Executor)

Planner + executor separation, tool selection guidance, sub-agent invocation.

Memory, State, and Long-Term Context

Memory layers in multi-step / multi-session agents: scratch, episodic, semantic, user profile.

Human-in-the-Loop and Approval Flows

Gating risky agent actions through human approval: pre-execution diff, severity tier, audit log.

Project: Multilingual Customer Support Assistant

TR/EN support pipeline: intent, FAQ, escalation, CSAT feedback. Which step on which model, and how to design evals.

Project: RAG Document Q&A System

RAG over company docs: chunking, embedding, retrieval, re-ranking, anchored answers.

Project: Code Review Automation

PR review automation with GitHub Actions + Claude: diff parsing, rubric-based comments, severity labels.

Project: Content Production Pipeline

From topic list to publish: outline → draft → editor → SEO → image suggestion → CMS publish. Human approval at each step.

Project: Structured Data Extraction from PDFs

Invoices, contracts, forms — structured extraction with Claude vision + tool use. Accuracy metrics and audit trail.

MCP (Model Context Protocol) Integration

What is MCP and why? How to connect local/remote MCP servers to Claude. Examples with Slack, Notion, Postgres.

Skills, Plugins, and Marketplace

What are Skills, when to write one, what plugins are, how the marketplace works. Examples and publishing your own.

Community, Documentation, and Continuous Learning

Anthropic docs, cookbooks, Discord, papers, and a daily routine for keeping up.

Capstone Project + Final Quiz + Certificate

Real-world synthesis: design, build, and run an agent end-to-end, then take the final certification exam.