How to AI
Intermediate Level
Another 100 Claude Words You Should Know
11 Downloads
21 Views
1 Pages
Aug 13, 2026
About This Cheat Sheet
Hey! I'm Saurabh Samir, and this is my comprehensive glossary of 100 essential Claude AI terms you must know in 2026. I've been deep in the Claude ecosystem for years, and I've compiled every key concept you'll encounter — from Essentials (About-me file, Thinking, Add-in, Agent, AI Fluency) to Advanced Concepts (Anti-AI file, API, Artifact, AskUserQuestion), Core Features (Browser use, Cache, Canvas, Chat), Platform Terms (Cowork, Cowork Project, Credits, Cursor), the 4D Framework (Delegation, Description, Discernment, Diligence), all the models (Haiku, Opus, Opus 4.8, Sonnet), tools (Gamma, Granola, MCP, Notion, Slack), and key concepts (Hallucination, LLM, Long context, Memory, Multimodal, Reasoning, Token). Whether you're a beginner learning Claude from the ground up or a power user brushing up on terminology, this cheat sheet has every term you need to know.
📚 Essentials (1-10)
- About-me file - Captures your voice for AI
- Thinking - Claude decides how hard to think
- Add-in - Claude in Excel/PPT sidebar
- Agent - AI that takes actions
- AI Fluency - Anthropic's flagship free course
- Anthropic - The company that builds Claude
- Anti-AI file - Strips AI tells from text
- API - Developer pipe into Claude
- Artifact - Document beside the chat
- AskUserQuestion - Claude interviews you
⚙️ Core Concepts (11-20)
- Browser use - Claude clicks around for you
- Cache - Stored context saves tokens
- Canvas - Side panel where Artifacts live
- Chat - Basic conversation window
- Claude 101 - Free intro certification
- Claude Code - Claude in your terminal
- Claude Design - claude.ai/design
- Computer use - Claude controls your desktop
- Connector - One-click bridge to an app
- Context - Everything Claude read in session
📁 Workspace (21-30)
- Context window - Max text Claude holds
- Cowork - Desktop agent reading your folder
- Cowork Project - Persistent memory across sessions
- Credits - Monthly token budget
- Cursor - AI code editor with Claude
- Custom instructions - Personal rules for every chat
- Cutoff - Last date of training data
- Delegation - "What to give AI" pillar
- Description - "How to brief AI" pillar
- Desktop app - Native Mac/Windows program
🧠 Framework (31-40)
- Diligence - "Verify the output" pillar
- Discernment - "Judge the answer" pillar
- Em-dash — Punctuation that screams "AI"
- Embedding - Numbers for finding similar text
- Enterprise - Plan for large companies
- Extended thinking - Claude reasons longer
- File-based - Workflows in folders, not chats
- Folder - Directory Cowork reads
- 4D Framework - Delegation, Description, Discernment, Diligence
- Frontier model - Newest, most capable AI
🔌 Integrations (41-50)
- Gamma - AI deck tool connector
- Granola - AI meeting-notes app
- Haiku - Claude's smallest, fastest model
- Hallucination - Confident answer that's wrong
- Input token - Text you send in
- Instructions - Rules for a whole project
- Knowledge - Files Claude always reads
- LLM - Large Language Model family
- Long context - Reads very long documents
- Lunch-and-learn - Doomed AI rollout method
📊 Models & Plans (61-70)
- Opus 4.8 - Current frontier Claude model
- Output token - Text Claude generates back
- Persona - Voice file making Claude sound like you
- Plan - Free, Pro, Max, Team, Enterprise
- Plugin - Skill or Connector
- Positive prompt - Tell what to do, not avoid
- Pro - $20/month paid plan
- Project - Folder of chats with shared files
- Prompt - Message you actually send
- Prompt template - Saved, reusable prompt
📚 100 Terms
Complete Glossary
Everything you need🧠 4D Framework
Delegation, Description
Discernment, Diligence🤖 4 Models
Opus, Sonnet, Haiku
Opus 4.8🔌 20+ Tools
Connectors & Integrations
Full ecosystem📚 Complete 100-Term Glossary
| # | Term | Simple Definition |
|---|---|---|
| 1 | About-me file | Text file that captures your voice for any AI |
| 2 | Thinking | Claude 4.8 deciding on its own how hard to think |
| 3 | Add-in | Claude living inside Excel or PowerPoint's sidebar |
| 4 | Agent | AI that takes actions, not just answers |
| 5 | AI Fluency | Anthropic's flagship free course on using AI well |
| 6 | Anthropic | The company that builds Claude |
| 7 | Anti-AI file | Markdown file that strips AI tells from text |
| 8 | API | Developer pipe into Claude, billed per token |
| 9 | Artifact | Document Claude builds beside the chat |
| 10 | AskUserQuestion | Claude's tool to interview you before acting |
| 11 | Browser use | Claude clicking around in a real browser for you |
| 12 | Cache | Stored context Claude reuses to save tokens |
| 13 | Canvas | Editable side panel where Artifacts live |
| 14 | Chat | Basic Claude conversation window |
| 15 | Claude 101 | Anthropic's free intro certification |
| 16 | Claude Code | Claude in your terminal, coding for you |
| 17 | Claude Design | Claude's design tool at claude.ai/design |
| 18 | Computer use | Claude controlling your desktop on its computer |
| 19 | Connector | One-click bridge between Claude and an app |
| 20 | Context | Everything Claude has read in current session |
| 21 | Context window | Maximum text Claude can hold at once |
| 22 | Cowork | Claude as a desktop agent reading your folder |
| 23 | Cowork Project | Cowork with persistent memory across sessions |
| 24 | Credits | Monthly token budget on your paid plan |
| 25 | Cursor | AI code editor running on Claude under the hood |
| 26 | Custom instructions | Personal rules saved across every chat |
| 27 | Cutoff | Last date Claude's training data covered |
| 28 | Delegation | "What to give AI" pillar of AI Fluency |
| 29 | Description | "How to brief AI" pillar of AI Fluency |
| 30 | Desktop app | Claude as a native Mac or Windows program |
| 31 | Diligence | "Verify the output" pillar of AI Fluency |
| 32 | Discernment | "Judge the answer" pillar of AI Fluency |
| 33 | Em-dash | Punctuation that screams "AI wrote this" (—) |
| 34 | Embedding | Numbers that let AI find similar text |
| 35 | Enterprise | Claude's plan built for large companies |
| 36 | Extended thinking | Claude reasoning longer before it replies |
| 37 | File-based | Workflows that live in folders, not chats |
| 38 | Folder | Directory Cowork reads before every session |
| 39 | 4D Framework | Delegation, Description, Discernment, Diligence |
| 40 | Frontier model | Newest, most capable AI on the market |
| 41 | Gamma | AI deck tool Claude can call as a connector |
| 42 | Granola | AI meeting-notes app that plugs into Claude |
| 43 | Haiku | Claude's smallest, fastest model |
| 44 | Hallucination | Confident answer that is simply wrong |
| 45 | Input token | Text you send in, counted toward limit |
| 46 | Instructions | Rules you set once for a whole project |
| 47 | Knowledge | Files in a Project Claude always reads |
| 48 | LLM | Large Language Model - AI family Claude belongs to |
| 49 | Long context | Claude's ability to read very long documents |
| 50 | Lunch-and-learn | Doomed way to roll out AI to a team |
| 51 | Markdown | Plain-text format (.md) Claude reads best |
| 52 | Max | Claude's $100-a-month personal plan |
| 53 | Maximum reasoning | Prompt trick that forces Claude to think hard |
| 54 | MCP | Model Context Protocol - standard for plugging apps in |
| 55 | Memory | Claude remembering you across sessions |
| 56 | Model | Version of Claude doing the work |
| 57 | Multimodal | AI that reads text, images, and PDFs together |
| 58 | Negative parallelism | "Not X, it's Y" pattern that flags AI text |
| 59 | Notion | Connector that turns Claude into a second brain |
| 60 | Opus | Claude's most powerful model family |
| 61 | Opus 4.8 | Current frontier Claude model |
| 62 | Output token | Text Claude generates back, also billed |
| 63 | Persona | Voice file that makes Claude sound like you |
| 64 | Plan | Free, Pro, Max, Team, or Enterprise tier |
| 65 | Plugin | Another word for a Skill or Connector |
| 66 | Positive prompt | Tell Claude what to do, never what to avoid |
| 67 | Pro | Claude's basic paid plan at $20/month |
| 68 | Project | Folder of chats with shared files and rules |
| 69 | Prompt | Message you actually send Claude |
| 70 | Prompt template | Saved, reusable prompt for a team |
| 71 | Rate limit | Cap on how much you can use Claude per window |
| 72 | Reasoning | Claude thinking step-by-step before answering |
| 73 | Refusal | When Claude declines a request on safety grounds |
| 74 | Rollout | Week-long plan to put a team on Claude |
| 75 | Scheduled task | Cowork running a job at a set time |
| 76 | Second brain | Your files becoming Claude's permanent memory |
| 77 | Send-ready | Output finished enough to ship with no edits |
| 78 | Send-to-Canvas | Claude Design button to push art into Canvas |
| 79 | Session | One continuous chat from open to close |
| 80 | Skill | Command that loads pre-baked instructions |
| 81 | Skilljar | Anthropic's free certification academy site |
| 82 | Slack | Connector that lets Claude read team channels |
| 83 | Slash command | Typing "/" to fire a Skill, like /47 |
| 84 | Sonnet | Claude's middle-tier balanced model |
| 85 | Style file | Text file teaching Claude your voice |
| 86 | Sycophancy | AI being so agreeable it stops being useful |
| 87 | System prompt | Hidden instructions running before yours |
| 88 | Taste | Human judgment AI still can't replace |
| 89 | Team | Claude's plan for 5 to 150 people |
| 90 | Token | Roughly one word; the unit Claude bills in |
| 91 | Tool | Something Claude can call, like web search or Gmail |
| 92 | Tool use | Claude choosing and running tools mid-task |
| 93 | Turn | One user message plus one Claude reply |
| 94 | Usage limit | Moment Claude says "you've reached your cap" |
| 95 | Vision | Claude reading images, screenshots, and PDFs |
| 96 | Voice | Writing personality you want Claude to copy |
| 97 | Wage premium | 56% salary bump workers with AI skills get |
| 98 | Web search | Claude looking things up on the internet live |
| 99 | Workspace | Shared Claude environment for a team |
| 100 | /47 | Skill that rewrites lazy prompts for Opus 4.8 |
📂 Category Breakdown
| Category | Terms | Key Concepts |
|---|---|---|
| Essential Basics | 1-10 | About-me, Thinking, Add-in, Agent, AI Fluency, Anthropic, Anti-AI, API, Artifact, AskUserQuestion |
| Core Features | 11-20 | Browser use, Cache, Canvas, Chat, Claude 101, Claude Code, Claude Design, Computer use, Connector, Context |
| Workspace & Context | 21-30 | Context window, Cowork, Cowork Project, Credits, Cursor, Custom instructions, Cutoff, Delegation, Description, Desktop app |
| AI Framework | 31-40 | Diligence, Discernment, Em-dash, Embedding, Enterprise, Extended thinking, File-based, Folder, 4D Framework, Frontier model |
| Integrations | 41-50 | Gamma, Granola, Haiku, Hallucination, Input token, Instructions, Knowledge, LLM, Long context, Lunch-and-learn |
| Advanced Features | 51-60 | Markdown, Max, Maximum reasoning, MCP, Memory, Model, Multimodal, Negative parallelism, Notion, Opus |
| Models & Plans | 61-70 | Opus 4.8, Output token, Persona, Plan, Plugin, Positive prompt, Pro, Project, Prompt, Prompt template |
| Usage & Limits | 71-80 | Rate limit, Reasoning, Refusal, Rollout, Scheduled task, Second brain, Send-ready, Send-to-Canvas, Session, Skill |
| Tools & Commands | 81-90 | Skilljar, Slack, Slash command, Sonnet, Style file, Sycophancy, System prompt, Taste, Team, Token |
| Advanced Concepts | 91-100 | Tool, Tool use, Turn, Usage limit, Vision, Voice, Wage premium, Web search, Workspace, /47 |
🧠 The 4D Framework — Key Terms
| Pillar | Term # | Definition | What It Means |
|---|---|---|---|
| Delegation | #28 | What to give AI | Decide what tasks to hand off |
| Description | #29 | How to brief AI | Write clear instructions |
| Discernment | #32 | Judge the answer | Evaluate if output is good |
| Diligence | #31 | Verify the output | Check for accuracy/errors |
🤖 Model Family Comparison
| Model | Term # | Description | Best For |
|---|---|---|---|
| Opus 4.8 | #61 | Current frontier Claude model | Most complex tasks |
| Opus | #60 | Most powerful model family | Deep reasoning |
| Sonnet | #84 | Middle-tier balanced model | Everyday tasks |
| Haiku | #43 | Smallest, fastest model | Quick responses |
📋 Plan Types
| Plan | Term # | Price | Best For |
|---|---|---|---|
| Free | #64 | $0 | Basic use |
| Pro | #67 | $20/month | Individual power users |
| Max | #52 | $100/month | Heavy personal use |
| Team | #89 | Custom | 5-150 people |
| Enterprise | #35 | Custom | Large companies |
🔌 Essential Tools & Connectors
| Tool | Term # | Purpose |
|---|---|---|
| AskUserQuestion | #10 | Interview you before acting |
| Gamma | #41 | AI deck tool |
| Granola | #42 | AI meeting-notes app |
| Notion | #59 | Second brain connector |
| Slack | #82 | Team channel connector |
| Web search | #98 | Live internet lookup |
| /47 | #100 | Rewrites lazy prompts for Opus 4.8 |
🧠 Key Concepts — Quick Reference
| Concept | Term # | Definition |
|---|---|---|
| Hallucination | #44 | Confident answer that's wrong |
| Sycophancy | #86 | AI being too agreeable |
| Refusal | #73 | Claude declining a request |
| Em-dash | #33 | Punctuation that screams "AI wrote this" (—) |
| Negative parallelism | #58 | "Not X, it's Y" pattern flags AI text |
| Wage premium | #97 | 56% salary bump with AI skills |
| Frontier model | #40 | Newest, most capable AI |
| Multimodal | #57 | Reads text, images, PDFs together |
📁 File Types
| File Type | Term # | Description |
|---|---|---|
| About-me file | #1 | Captures your voice for AI |
| Anti-AI file | #7 | Strips AI tells from text |
| Style file | #85 | Teaches Claude your voice |
| Markdown | #51 | Plain-text format (.md) Claude reads best |
| Persona | #63 | Voice file for Claude to sound like you |
📊 Usage Terms
| Term | # | Definition |
|---|---|---|
| Token | #90 | Roughly one word; billing unit |
| Input token | #45 | Text you send in |
| Output token | #62 | Text Claude generates back |
| Context window | #21 | Maximum text Claude can hold |
| Rate limit | #71 | Cap on usage per window |
| Usage limit | #94 | Reached your cap message |
| Session | #79 | One chat from open to close |
| Turn | #93 | One user message + one reply |
🎓 Certification & Learning
| Resource | Term # | Description |
|---|---|---|
| AI Fluency | #5 | Anthropic's flagship free course |
| Claude 101 | #15 | Free intro certification |
| Skilljar | #81 | Free certification academy site |
| 4D Framework | #39 | Delegation, Description, Discernment, Diligence |
🎯 Another 100 Claude Words — Quick Revision
The Golden Rule: Understanding these 100 terms will make you fluent in the Claude ecosystem. From the 4D Framework to token billing, each term represents a piece of the puzzle that makes Claude the most powerful AI assistant available.
Quick Memory Tricks:
- 📌 4D Framework: "Delegation (what to give), Description (how to brief), Discernment (judge), Diligence (verify)"
- 📌 3 Models: "Opus (smartest), Sonnet (balanced), Haiku (fastest)"
- 📌 5 Plans: "Free → Pro ($20) → Max ($100) → Team → Enterprise"
- 📌 Key Terms: "Hallucination (wrong), Sycophancy (too agreeable), Refusal (declines)"
- 📌 The Trap: "Em-dash (—) and Negative parallelism ('Not X, it's Y') scream 'AI wrote this'"
Must-Know Terms:
- ✅ /47 — Skill that rewrites lazy prompts for Opus 4.8
- ✅ AskUserQuestion — Claude interviews you before acting
- ✅ MCP — Model Context Protocol, standard for plugging apps in
- ✅ Wage premium — 56% salary bump with AI skills
- ✅ Context window — Maximum text Claude can hold at once
💡 Pro Tips for Learning These Terms
✅ Learn by Category
Don't memorize all 100 at once. Learn by category — start with Essentials, then Core Features, then Models & Plans.
✅ Use the 4D Framework
The 4D Framework (Delegation, Description, Discernment, Diligence) is the foundation of AI Fluency. Master these first.
✅ Understand Token Billing
Tokens are how Claude bills you. Input tokens (what you send) + Output tokens (what you get back) = your total cost.
✅ Know the Models
Opus = complex reasoning, Sonnet = everyday tasks, Haiku = quick answers. Pick the right model for each task.
✅ Learn the Flags
Em-dash (—) and Negative parallelism ('Not X, it's Y') are patterns that scream "AI wrote this." Learn to spot and avoid them.
✅ Use /47
The /47 skill rewrites lazy prompts for Opus 4.8. It's one of the most powerful shortcuts in the Claude ecosystem.
from Saurabh Samir, on "How to AI"
Download Cheat Sheet
Click the button below to download this high-resolution PNG cheat sheet.
Download PNG (1.5 MB)Preview
Still have questions?
We're here to help! Reach out to our support team for any queries.