Skip to content

Capability Map

The capability map is not another curriculum, but a professional view of the same content. It treats AI Coding as a complete engineering system and helps learners, platform builders, and technical leaders identify the problems, boundaries, and governance methods at every layer.

Two-tier professional model 

Life cycle layer: input and context → Agent execution and capability expansion → output and verification
Governance layer: Model management · Gateway management · Cost analysis · Quality assessment
Across the entire life cycle

The first three items describe how an AI Coding task flows; the last four items are not subsequent steps, but governance dimensions that act on the entire process of input, execution, and output.

On top of these two foundational layers, specialization tracks reorganize knowledge across several dimensions into a continuous engineering path.

Input and context

Task lifecycle

Focus on whether AI receives correct, sufficient, and clearly bounded information.

  • Project rules: from Rules to AGENTS.md
  • Requirements: from natural language to specification-driven development
  • Context Engineering: selecting, compressing, and updating context in large codebases

Core articles

Intermediate stage24 minutes Let AI Coding Agent truly understand the project: from Rules to AGENTS.mdTurn project conventions into rules that Agents can continuously read, and reduce repeated mistakes through hierarchical scopes, real commands, and validation requirements. Intermediate Stage28 minutes Why Does AI Coding Get Messier with Every Change? From Requirements to Spec-Driven DevelopmentTurn ambiguous requests into bounded, traceable, and verifiable specifications so AI Coding can move from repeated guesswork to reliable delivery Advanced Stage26 minutes Context Engineering: How AI Coding Agents Find Their Way Through a Large CodebaseSee how coding agents search, index, map, and verify a repository, then learn a practical way to give them less noise and better evidence
Novice stage10 minutes Quick Start with AI Coding: Develop a web application in 10 minutesComplete the first end-to-end practice of AI Coding through a small web application Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Execution and capability expansion

Task lifecycle

Focus on how Agents invoke tools, connect systems, reuse experience, and complete complex tasks.

  • MCP, Tools, and external systems
  • From Prompt to Agent Skill
  • From Rules and Skills to Plugins
  • Complex task decomposition and multi-agent workflows

Core articles

Beginner Stage30 minutes Install and Compare Mainstream AI Coding ToolsA practical beginner's map of six representative AI Coding tools: how visual editors, terminal agents, and conversational autonomous agents differ, what to install, and how to choose safely. Novice stage10 minutes Quick Start with AI Coding: Develop a web application in 10 minutesComplete the first end-to-end practice of AI Coding through a small web application Beginner Stage25 minutes From Localhost to a Global Website: Deploy with Codex, GitHub, and CloudflareA beginner-friendly, hands-on guide to build and verify with ChatGPT Codex, manage code on GitHub, and continuously deploy a web app with Cloudflare Pages Intermediate stage32 minutes MCP Panorama Guide: How AI can securely connect the real worldFrom life metaphors to protocol messages, the system understands MCP, Tools, Host, Client, Server, permissions and security boundaries Advanced Stage28 minutes From Prompt to Agent Skill: Package Expert Knowledge as a Reusable CapabilityTurn a one-off prompt into a focused, testable, maintainable Agent Skill with activation rules, workflows, resources, and evaluations Advanced Stage35 minutes From Rules and Skills to Plugins: Engineering Composable Agent CapabilitiesCompare Cursor, Claude Code, and Codex plugin systems, understand lifecycle hooks, and build an installable capability package with portable skills and host-specific adapters Advanced Stage24 minutes Multi-Agent Workflow for Complex AI Coding TasksUnderstand when one coding Agent is enough, when several Agents actually help, and how to divide, hand off, integrate, and verify complex work Advanced Stage26 minutes Multica Multi-Agent Work ModeUnderstand Multica's operating model, migrate an existing AGENTS.md + rules/skills workflow onto it without breaking anything, and run reliable single-project and team-level multi-agent collaboration
Beginner Stage16 minutes How to Choose an AI Coding Tool: A Plain-Language Guide for BeginnersStop memorizing product names and chasing rankings. Learn how to choose among code completion, chat and planning, IDE agents, and terminal or cloud agents based on the task. Intermediate stage24 minutes Let AI Coding Agent truly understand the project: from Rules to AGENTS.mdTurn project conventions into rules that Agents can continuously read, and reduce repeated mistakes through hierarchical scopes, real commands, and validation requirements. Intermediate Stage28 minutes Why Does AI Coding Get Messier with Every Change? From Requirements to Spec-Driven DevelopmentTurn ambiguous requests into bounded, traceable, and verifiable specifications so AI Coding can move from repeated guesswork to reliable delivery Intermediate stage24 minutes Let the AI ​​Coding Agent open the browser and troubleshoot by itself: Chrome DevTools MCP practiceConnect Chrome DevTools MCP in Cursor, Claude Code and Codex, let the Agent reproduce front-end problems, read Console and Network, repair and regression verification Advanced Stage26 minutes Context Engineering: How AI Coding Agents Find Their Way Through a Large CodebaseSee how coding agents search, index, map, and verify a repository, then learn a practical way to give them less noise and better evidence Master Stage24 minutes Build a Local Control Plane with CC SwitchTurn scattered provider, credential, model, MCP, prompt, and Skill settings into inspectable local profiles—and understand where CC Switch stops and a real AI Gateway begins. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Output and verification

Task lifecycle

Focus on whether generated results are correct, maintainable, and verifiable.

  • From code generation to verification engineering
  • Testing, static analysis, review, and delivery evidence
  • Change scope, regression risk, and quality gates

Core articles

Intermediate Stage26 minutes How Can AI Coding Maintain Code Quality? From Code Generation to Verification EngineeringBuild a risk-based verification loop with TDD, layered tests, AI-assisted review, end-to-end evidence, and enforceable quality gates Intermediate stage24 minutes Let the AI ​​Coding Agent open the browser and troubleshoot by itself: Chrome DevTools MCP practiceConnect Chrome DevTools MCP in Cursor, Claude Code and Codex, let the Agent reproduce front-end problems, read Console and Network, repair and regression verification
Beginner Stage16 minutes How to Choose an AI Coding Tool: A Plain-Language Guide for BeginnersStop memorizing product names and chasing rankings. Learn how to choose among code completion, chat and planning, IDE agents, and terminal or cloud agents based on the task. Novice stage10 minutes Quick Start with AI Coding: Develop a web application in 10 minutesComplete the first end-to-end practice of AI Coding through a small web application Beginner Stage25 minutes From Localhost to a Global Website: Deploy with Codex, GitHub, and CloudflareA beginner-friendly, hands-on guide to build and verify with ChatGPT Codex, manage code on GitHub, and continuously deploy a web app with Cloudflare Pages Intermediate stage24 minutes Let AI Coding Agent truly understand the project: from Rules to AGENTS.mdTurn project conventions into rules that Agents can continuously read, and reduce repeated mistakes through hierarchical scopes, real commands, and validation requirements. Intermediate Stage28 minutes Why Does AI Coding Get Messier with Every Change? From Requirements to Spec-Driven DevelopmentTurn ambiguous requests into bounded, traceable, and verifiable specifications so AI Coding can move from repeated guesswork to reliable delivery Advanced Stage26 minutes Context Engineering: How AI Coding Agents Find Their Way Through a Large CodebaseSee how coding agents search, index, map, and verify a repository, then learn a practical way to give them less noise and better evidence Advanced Stage24 minutes Multi-Agent Workflow for Complex AI Coding TasksUnderstand when one coding Agent is enough, when several Agents actually help, and how to divide, hand off, integrate, and verify complex work Advanced Stage26 minutes Multica Multi-Agent Work ModeUnderstand Multica's operating model, migrate an existing AGENTS.md + rules/skills workflow onto it without breaking anything, and run reliable single-project and team-level multi-agent collaboration Master Stage20 minutes Understand AI Coding Token UsageBuild a practical mental model of AI Coding token consumption, read the native usage views in Claude Code, Codex, and Cursor, and use ccusage only when local history answers the next question better. Master Stage20 minutes AI Coding Token and Cost AnalysisA practical look at what an AI Coding Agent reads, calls, retries, and hands back to a reviewer during one engineering task. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Horizontal governance layer

Model management

Cross-cutting governance

Focus on model selection, routing, capability boundaries, and real engineering performance.

  • Model capability boundaries and task fit
  • Benchmarks and real-project evaluation
  • Model version changes and regression verification

Core articles

Beginner Stage16 minutes How to Choose an AI Coding Tool: A Plain-Language Guide for BeginnersStop memorizing product names and chasing rankings. Learn how to choose among code completion, chat and planning, IDE agents, and terminal or cloud agents based on the task. Master Stage22 minutes Understand BYOK and API KeysStart with the relationship between the AI coding product layer and underlying model calls, then understand product-managed subscriptions, BYOK, billing, data paths, and security responsibilities. Master Stage24 minutes Build a Local Control Plane with CC SwitchTurn scattered provider, credential, model, MCP, prompt, and Skill settings into inspectable local profiles—and understand where CC Switch stops and a real AI Gateway begins.
Master Stage20 minutes Understand AI Coding Token UsageBuild a practical mental model of AI Coding token consumption, read the native usage views in Claude Code, Codex, and Cursor, and use ccusage only when local history answers the next question better. Master Stage20 minutes AI Coding Token and Cost AnalysisA practical look at what an AI Coding Agent reads, calls, retries, and hands back to a reviewer during one engineering task. Master Stage26 minutes From API Relays to AI GatewaysSee what really happens between an AI Coding client and the model API: authentication, account pools, routing, fallback, metering, audit—and the risks of an opaque relay. Master Stage24 minutes From Benchmarks to a Real Project TrialUnderstand what coding benchmarks actually test, where to find credible live leaderboards, and how to compare model routes with a small set of real project tasks. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Gateway management

Cross-cutting governance

Focus on a unified entry point for model calls, permissions, rate limits, routing, and auditing.

  • AI Gateway responsibilities and boundaries
  • Multi-model routing, fallback, and resilience
  • Permissions, quotas, and call auditing

Core articles

Master Stage26 minutes From API Relays to AI GatewaysSee what really happens between an AI Coding client and the model API: authentication, account pools, routing, fallback, metering, audit—and the risks of an opaque relay.
Master Stage24 minutes Build a Local Control Plane with CC SwitchTurn scattered provider, credential, model, MCP, prompt, and Skill settings into inspectable local profiles—and understand where CC Switch stops and a real AI Gateway begins. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Cost analysis

Cross-cutting governance

Focus on the real cost of tokens, tool calls, retries, and context strategies.

  • Claude Code and Codex usage analysis
  • Task-level cost attribution
  • Trade-offs among quality, speed, and cost

Core articles

Master Stage20 minutes Understand AI Coding Token UsageBuild a practical mental model of AI Coding token consumption, read the native usage views in Claude Code, Codex, and Cursor, and use ccusage only when local history answers the next question better. Master Stage20 minutes AI Coding Token and Cost AnalysisA practical look at what an AI Coding Agent reads, calls, retries, and hands back to a reviewer during one engineering task.
Master Stage22 minutes Understand BYOK and API KeysStart with the relationship between the AI coding product layer and underlying model calls, then understand product-managed subscriptions, BYOK, billing, data paths, and security responsibilities. Master Stage26 minutes From API Relays to AI GatewaysSee what really happens between an AI Coding client and the model API: authentication, account pools, routing, fallback, metering, audit—and the risks of an opaque relay. Master Stage24 minutes From Benchmarks to a Real Project TrialUnderstand what coding benchmarks actually test, where to find credible live leaderboards, and how to compare model routes with a small set of real project tasks. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.

Quality assessment

Cross-cutting governance

Focus on how teams continuously measure whether AI Coding is genuinely effective.

  • Code correctness and maintainability
  • End-to-end task success rate
  • Efficiency gains, rework rate, and human intervention
  • Team-level evaluation mechanisms

Core articles

Master Stage24 minutes From Benchmarks to a Real Project TrialUnderstand what coding benchmarks actually test, where to find credible live leaderboards, and how to compare model routes with a small set of real project tasks. Master Stage32 minutes Build a Sustainable Team AI Coding SystemBuild one maintainable AI Coding operating system for mixed-skill teammates and multiple tools: canonical project docs, thin tool adapters, task contracts, executable guardrails, and a documentation feedback loop. Master Stage30 minutes AI Coding Project Lifecycle ManagementAdapt a proven AI project lifecycle methodology — Exploration, Mobilization, Execution, Delivery — for teams using AI coding agents. Includes cost estimation, risk management, iteration control, and a practical readiness checklist.
Intermediate Stage26 minutes How Can AI Coding Maintain Code Quality? From Code Generation to Verification EngineeringBuild a risk-based verification loop with TDD, layered tests, AI-assisted review, end-to-end evidence, and enforceable quality gates Advanced Stage28 minutes From Prompt to Agent Skill: Package Expert Knowledge as a Reusable CapabilityTurn a one-off prompt into a focused, testable, maintainable Agent Skill with activation rules, workflows, resources, and evaluations

Specialization Tracks

SPECIALIZATION 01 Agent Engineering Move from using and configuring agents to understanding the agent runtime and building agent applications that are observable, evaluable, and safe to operate. Agent Loop · Tool Calling · Context · Sandbox · Orchestration · Evaluation Explore the track →

How to coordinate the two routes

Follow the Growth Route when learning for the first time. Use this page to inspect a professional concern when designing a platform, troubleshooting a system, or building team standards. When you want to understand and implement agent systems, continue with the Agent Engineering track.