Reference · module 6 of 6
Official links
Where to check when this track goes stale — which it will.
Start here
| Claude Platform docs | https://platform.claude.com/docs |
| Claude Code docs | https://code.claude.com/docs |
| Help Center (consumer features) | https://support.claude.com |
| Console | https://platform.claude.com |
| claude.ai | https://claude.ai |
Machine-readable indexes — useful if you’re pointing Claude at the docs:
- https://code.claude.com/docs/llms.txt
- https://platform.claude.com/llms.txt
What changes fastest
| Check | For |
|---|---|
| Claude Code What’s New | Weekly digest of new features |
| Claude Code changelog | Version-by-version |
| Platform release notes | API changes |
| Help Center release notes | Consumer features |
| Models overview | Model names, prices, capabilities |
| Model deprecations | What’s being retired |
| Beta headers | What still needs one |
| Available beta and research preview features | What’s in preview |
By topic
Prompting
- Prompting best practices — the canonical reference
- Prompt engineering overview
- Per-model guides: Opus 5 · Sonnet 5 · Fable 5 · Opus 4.8
- Prompt library (Claude Code)
Models and pricing
Core API
- Quickstart
- Using the Messages API
- Thinking · Effort
- Structured outputs
- Prompt caching
- Streaming · Batch
- Vision · PDFs · Files API
- Citations
- Context windows · Compaction · Context editing
- Errors · Rate limits
Tools
- Tool use overview
- Define tools · Handle tool calls
- Server tools
- Tool search · Programmatic tool calling
- Tool Runner
- Build a tool-using agent
Skills
- Agent Skills overview
- Authoring best practices
- Skills in the API
- Skills in Claude Code
- Skills for enterprise
- anthropics/skills · agentskills.io
MCP
- MCP in Claude Code · quickstart
- MCP connector (API)
- Remote MCP servers
- MCP tunnels
- Managed MCP
- modelcontextprotocol.io · Connectors directory
Claude Code
- Overview · Quickstart
- How Claude Code works
- Extend Claude Code — when to use what
- Best practices · Common workflows
- CLI reference · Commands · Settings
- Memory / CLAUDE.md
- Hooks guide · Hooks reference
- Subagents · Agent teams · Agents comparison · Dynamic workflows
- Permissions · Permission modes · Auto mode config
- Sandboxing · Sandbox environments
- Plugins · Marketplaces · Discover plugins
- Costs · Context window (interactive)
- Headless · GitHub Actions · GitLab CI/CD
- Routines
- Debug your config · Troubleshooting · Errors
- Glossary
Agent SDK
- Overview · Quickstart
- Python · TypeScript
- Agent loop · Custom tools
- Permissions · Hooks · Subagents
- Sessions · Session storage
- Structured outputs · Cost tracking
- Hosting · Secure deployment · Observability
Managed Agents
Testing and safety
- Define success and build evaluations
- Mitigate jailbreaks and prompt injections
- Reduce hallucinations
- Increase output consistency
- Reduce prompt leak
- Reducing latency
- Security (Claude Code)
Enterprise
- Admin setup (Claude Code)
- Admin API · Workspaces
- Workload Identity Federation
- Compliance API
- Data retention · Zero data retention
- Feature availability
- Trust Center
Consumer features
- Projects · RAG for projects
- Skills · Creating custom skills
- Cowork · Using Cowork safely
- Artifacts
- File creation
- Research
- Memory and chat search
- Skills, connectors and plugins directory
- Provision and manage skills for your organization
- Claude Code cheatsheet (Help Center)
- Usage limits