Seekvana
AdvancedNo experience neededFree

Inside the Agent

from first tool call to autonomous agent

120 hands-on articles across 13 modules — build real AI agents inside Cursor and Claude Code, the tools professionals actually use. Tool use, RAG, MCP, context engineering, memory, the Claude Agent SDK, multi-agent orchestration, evals, security, and production deployment — every lesson read, built, and shipped in a real editor, ending in benchmarked capstones.

13modules
120topics
120tasks
25–35 hrstotal
Start Module 01View all paths

Curriculum

13 modules · 120 topics

Every article runs concept → build → ship — the idea, a hands-on lab in Cursor or Claude Code, and a graded task you keep. Where a lesson needs a prerequisite it links to Getting Started or Beyond the Prompt instead of repeating it.

  • 14.01What This Path Is — and What It Assumes
  • 14.02Installing and Touring Cursor
  • 14.03Installing Claude Code
  • 14.04Cursor's Modes and Inline Edits
  • 14.05@-Context and Codebase Indexing
  • 14.06Claude Code as an Agentic Terminal
  • 14.07The Coding-Agent Landscape (2026)
  • 14.08Safety-First Defaults
Each topic includes a hands-on lab8 tasks
  • 15.01Prompting vs Agentic Workflows
  • 15.02The Augmented LLM
  • 15.03What Is an Agent, Really?
  • 15.04Anatomy of an Agent and the Think → Act → Observe Loop
  • 15.05When NOT to Build an Agent
  • 15.06Autonomy as a Spectrum
  • 15.07Your First Agent, Felt Not Coded
Each topic includes a hands-on lab7 tasks
  • 16.01Chain-of-Thought, Revisited for Agents
  • 16.02ReAct: Reason + Act
  • 16.03Plan-and-Execute
  • 16.04Reflection and Self-Critique
  • 16.05The Evaluate-Improve Loop
  • 16.06Reasoning Models and Test-Time Compute
  • 16.07Agent Skills, Roles, and Capabilities
  • 16.08Spec-Driven Development
  • 16.09Decision Journaling
Each topic includes a hands-on lab9 tasks
  • 17.01How Tool and Function Calling Actually Works
  • 17.02Designing Good Tools
  • 17.03CodeAct vs JSON Tool Calls
  • 17.04Calling External APIs from an Agent
  • 17.05Querying Databases from an Agent
  • 17.06Structured Outputs and Schema Validation
  • 17.07Embeddings, Vector Stores, and Why RAG
  • 17.08Building a RAG Pipeline in Cursor
  • 17.09Agentic RAG
  • 17.10Evaluating Retrieval
Each topic includes a hands-on lab10 tasks
  • 18.01Why MCP Exists
  • 18.02MCP Architecture
  • 18.03The Three Primitives: Tools, Resources, Prompts
  • 18.04Adding MCP Servers to Cursor
  • 18.05Adding MCP Servers to Claude Code
  • 18.06Using Prebuilt Servers
  • 18.07Building Your First MCP Server with FastMCP
  • 18.08Building an MCP Client
  • 18.09Deploying a Remote MCP Server
  • 18.10MCP Authorization: OAuth 2.1 and PKCE
  • 18.11Code Execution with MCP
  • 18.12MCP Security and the Trust Boundary
Each topic includes a hands-on lab12 tasks
  • 19.01The Statelessness Problem
  • 19.02Context Rot and "Lost in the Middle"
  • 19.03Compaction and Summarization
  • 19.04Structured Note-Taking and Tool-Result Pruning
  • 19.05Just-in-Time vs Pre-Loaded Context
  • 19.06Memory via Files: CLAUDE.md
  • 19.07The CLAUDE.md Hierarchy and @-Imports
  • 19.08Cursor Rules and AGENTS.md
  • 19.09Progress Files and Shift Logs
  • 19.10Database and Vector-Store Memory
  • 19.11Memory Research: MemGPT/Letta and Sleep-Time Compute
Each topic includes a hands-on lab11 tasks
  • 20.01The Extensibility Stack, Mapped
  • 20.02Slash Commands
  • 20.03Skills: Authoring and Auto-Triggering
  • 20.04Subagents
  • 20.05Hooks: Deterministic Control
  • 20.06The Plugin System and Marketplaces
  • 20.07Essential Plugins, Toured
  • 20.08Plugin Safety
  • 20.09Building and Publishing Your Own Plugin
  • 20.10Plan Mode and Multi-Context-Window Workflows
  • 20.11The Initializer / Harness Pattern
  • 20.12Failure Modes, Fixes, and Cursor Parallels
Each topic includes a hands-on lab12 tasks
  • 21.01Agent SDK vs the Messages API
  • 21.02Install and First Query
  • 21.03The Agent Loop Under the Hood
  • 21.04Configuring the Agent
  • 21.05Choosing Your Model
  • 21.06Built-in Tools
  • 21.07Custom Tools and In-Process MCP
  • 21.08Streaming and Sessions
  • 21.09Checkpointing and Permissions
  • 21.10Cost, Caching, and Reliability Engineering
  • 21.11Harness Thinking and the No-Code Counterpoint
Each topic includes a hands-on lab11 tasks
  • 22.01Multi-Step Planning
  • 22.02The Five Workflow Patterns
  • 22.03Orchestration vs Choreography
  • 22.04When Multi-Agent Breaks
  • 22.05Subagent Orchestration in Claude Code
  • 22.06LangGraph
  • 22.07CrewAI and AutoGen
  • 22.08Choosing Your Orchestration Approach
Each topic includes a hands-on lab8 tasks
  • 23.01Why Evals First
  • 23.02Objective vs Subjective Evals
  • 23.03Response, Step, and Trajectory Evaluation
  • 23.04pass@k vs pass^k
  • 23.05Error Analysis with Traces
  • 23.06Eval Platforms and Benchmarks
  • 23.07The Epistemics of Benchmarks
  • 23.08Guardrails and Bounded Autonomy
  • 23.09Deterministic Enforcement
  • 23.10Prompt Injection and the Lethal Trifecta
  • 23.11Human-in-the-Loop
  • 23.12Responsible AI in Practice
Each topic includes a hands-on lab12 tasks
  • 24.01From Notebook to Service
  • 24.02Containerizing Agents with Docker
  • 24.03CI/CD for Agents
  • 24.04Observability and Tracing
  • 24.05Cost and Latency Management
  • 24.06State and Checkpointing at Scale
  • 24.07Hosting the Agent SDK Securely
  • 24.08Scaling, Enterprise Integration, and AgentOps Maturity
Each topic includes a hands-on lab8 tasks
  • 25.01MCP Governance and the Registry
  • 25.02AGENTS.md as a Cross-Tool Standard
Each topic includes a hands-on lab2 tasks
  • 26.01Watching the Agentic Web: A2A and Agent Commerce
  • 26.02Computer-Use and Browser Agents
  • 26.03Multimodal Agents
  • 26.04Human-Agent Collaboration
  • 26.05Application Patterns Across Verticals
  • 26.06Production Case Studies and the ROI Reality
  • 26.07Capstone A: The Deep-Research Agent
  • 26.08Capstone B: The OAuth-Secured MCP Platform
  • 26.09Capstone C: The Autonomous Feature Builder
  • 26.10Capstone D: Your Own Agent and Portfolio
Each topic includes a hands-on lab10 tasks

Frequently asked questions