The complete feature surface of 📝Claude, organized by product surface rather than alphabetically and compiled from 📝Anthropic's published documentation. Features with a dedicated memo link to it; the rest are listed by capability. 📝Claude Code carries its own index at 📝Claude Code Features. For the full collection, browse 🏷️#feature-of-claude.

Claude Apps — Chat and Content

📝Memory — generates entries from your chats and groups them by category so context carries across conversations
Artifacts — a side panel where Claude builds documents, code, and interactive apps you can edit, run, and share
Projects — a container with its own knowledge, instructions, and separate memory space for one body of work
Research — multi-step investigation across the web and connected apps that returns a cited report
Web search — current information from the web with inline citations when an answer sits past the model's knowledge
Extended thinking — visible step-by-step reasoning before the final answer on models that support it
File uploads — PDFs, images, spreadsheets, and documents read directly into the conversation
File creation and editing — produces and revises real Excel, Word, PowerPoint, and PDF files you can download
Skills — packaged instructions and scripts Claude loads on demand when a task matches
Custom skills — your own instructions and resources, uploaded and invoked the same way as Anthropic's
Plugins — bundles that extend Claude with additional capabilities from a single install
Retrieval augmented generation (RAG) — grounds answers in your own uploaded and connected content
Visual and interactive content — diagrams, charts, and interactive views rendered inline in the response
Memory import/export — moves memory between Claude and other AI services
Monthly recap — a periodic summary of how you have been using Claude
Claude Security — security-focused workflows surfaced inside the Claude app
📝Claude Design — a design surface for building and iterating on interfaces with Claude

Claude Apps — Personalization and Limits

Model selection — choose which Claude model answers, per conversation
Effort settings — trade response thoroughness against token efficiency
Thinking settings — control how much Claude reasons before answering
Appearance customization — adjust how the Claude interface looks
Language preferences — set the language Claude responds in
Break reminders — periodic nudges to step away during long sessions
Quiet hours — windows where Claude holds notifications
Usage limit management — see and manage your consumption against plan limits
Usage bundles — additional usage purchased on top of a plan
Length limits — caps on how long a single conversation or response can run

Claude in Your Workspace

Claude for Excel — works inside spreadsheets to build, analyze, and explain models
Claude for Word — drafts and revises documents in place
Claude for PowerPoint — builds and edits decks from your source material
Claude for Outlook — drafts, summarizes, and triages mail
Microsoft 365 integration — connects Claude across the Microsoft 365 suite
Claude in Xcode — brings Claude into Apple's IDE for native development

Claude Cowork

📝Claude Cowork — autonomous multi-step knowledge work that returns finished deliverables rather than chat replies
Local file access — reads and writes files on your machine, citing the actual files it used
Connected apps — works across services like Slack and Google Drive inside a single task
Scheduled tasks — runs a Cowork task on a daily, weekly, or monthly cadence
Web and mobile handoff — start a task at your desk and check on it from a browser or phone, in beta

Claude Developer Platform — Model Capabilities

📝Context windows — up to 1M tokens for large documents, extensive codebases, and long conversations
Adaptive thinking — Claude decides when and how much to think; the only thinking mode on 4.7 and later
Effort — controls how many tokens Claude spends responding, trading thoroughness against efficiency
Structured outputs — guarantees schema conformance via JSON outputs and strict tool use
Citations — grounds responses in source documents with references to the exact sentences used
Search results — natural citations for RAG applications over your own knowledge bases
PDF support — processes text and visual content from PDF documents
Batch processing — asynchronous processing of large request volumes at 50% of standard cost
Data residency — routes inference globally or to the US per request
Server-side fallback — retries a refused request against other models inside a single API call, in beta
Fallback credit — avoids paying the prompt-cache cost twice when retrying a refused request, in beta

Claude Developer Platform — Tools

Web search — augments Claude's knowledge with current data from across the web
Web fetch — retrieves full content from specified web pages and PDFs for in-depth analysis
Code execution — runs code in a sandbox for data analysis, calculation, and file processing
Advisor tool — pairs a fast executor model with a higher-intelligence advisor for long-horizon work, in beta
Bash — executes shell commands and scripts on your side of the call
Text editor — creates and edits text files through a built-in editor interface
Memory tool — stores and retrieves information across conversations to build knowledge over time
Computer use — controls interfaces by taking screenshots and issuing mouse and keyboard commands, in beta

Claude Developer Platform — Tool Infrastructure

Agent Skills — pre-built and custom skills using progressive disclosure to manage context efficiently, in beta
MCP connector — connects to remote MCP servers straight from the Messages API, in beta
Tool search — scales to thousands of tools by discovering and loading them on demand
Programmatic tool calling — lets Claude call your tools from inside code execution, cutting latency and tokens
Fine-grained tool streaming — streams tool parameters without buffering, reducing latency on large inputs

Claude Developer Platform — Context Management

Automatic prompt caching — caches the last cacheable block and moves the cache point forward as you go
Prompt caching — reuses background knowledge and examples across requests to cut cost and latency
📝ENABLE_PROMPT_CACHING_1H — the extended one-hour cache for important but less frequently hit context
Compaction — server-side summarization of earlier turns as a conversation approaches the window limit, in beta
Context editing — configurable strategies for clearing tool results and managing thinking blocks, in beta
Token counting — counts tokens in a message before you send it

Claude Developer Platform — Files and Assets

Files API — upload and manage PDFs, images, and text files once instead of resending them per request, in beta

Claude Managed Agents

📝Claude Managed Agents — server-hosted agents that run the loop and the sandbox for you
📝Agent Loop — the managed execution cycle that runs an agent's turns to completion
Cloud Containers — the default managed execution environment for a hosted agent
📝Self-Hosted Sandboxes — run the agent's execution environment on your own infrastructure
📝MCP Tunnels — expose a local or private MCP server to a hosted agent

Models

📝Claude Opus — Opus 5 for complex agentic coding and enterprise work, at a 1M-token context window
Claude Sonnet 5 — the best combination of speed and intelligence, also at 1M tokens
Claude Haiku 4.5 — the fastest model with near-frontier intelligence, at 200k tokens
📝Claude Fable 5 — next-generation intelligence for long-running agents, with adaptive thinking always on
📝Claude Mythos 5 — Fable 5's specs for defensive cybersecurity, invitation-only under 📝Project Glasswing

Deployment and Administration

Claude Platform on AWS — Anthropic-operated deployment using first-party model IDs and lifecycle
Amazon Bedrock — AWS-operated access to Claude through the Bedrock endpoint
Google Cloud — Google-operated access through Vertex AI
Claude in Microsoft Foundry — Anthropic-operated deployment on Azure, with availability varying by hosting option
Admin API — programmatic administration of an organization's workspaces, members, and keys
Usage and Cost API — programmatic reporting on consumption and spend
Compliance API — programmatic access to compliance and governance records
Zero Data Retention — per-feature eligibility for arrangements where no prompt or output content is stored