Skip to main content

neural-forge.io

Learn Tracks Models AI Explorer Compare

Sign inStartStart learning

Tendril

Tendril neural-forge.io

Free AI literacy for everyone, supported by trust-safe partners.

Learn

Curriculum
Tracks
For you
Preferences

Resources

Glossary
In the Wild
Newsroom
Community
Partners
Send Feedback
Changelog
About
New to AI?

Schools & Orgs

Schools
Libraries
Tech Teams
Free Access
Sponsor
Sign Up
Support the Mission

Trust

Privacy
Terms
COPPA
Accessibility

Legal

Privacy
Terms
COPPA
Accessibility

© 2026 Tendril·Privacy·Terms·Contact

Built with Claude

Loading lesson…

Tendril

Tools Literacy0%

Time on lesson

0s

← Tools Literacy

0 of 354 complete

○Lesson 391Subscription-Tier Literacy: Every Plan, Side by Side
○Lesson 392When to Upgrade (And When Not To)
○Lesson 393API Access vs. Consumer Products — A Deeper Look
○Lesson 394Building a Personal AI Stack for School and Career
○Lesson 395Projects and Spaces — Persistent Context Is the Future
○Lesson 396Privacy Settings Across the Big Three
○Lesson 397Tool Switching — Why You Shouldn't Marry One Model
○Lesson 401Perplexity Comet — the AI browser
○Lesson 671Cursor: The AI Code Editor That Ate Enterprise
○Lesson 672Windsurf: The Cursor Challenger With An Agent-First Vision
○Lesson 673Claude Code: Anthropic's Terminal-Native Coding Agent
○Lesson 674Codex CLI: OpenAI's Answer to Claude Code
○Lesson 675Zed: The Editor Built For AI From The Start
○Lesson 676Figma AI: When Design Tools Started Designing Themselves
○Lesson 677Framer AI: Design, Code, And Ship A Website In One Prompt
○Lesson 678Recraft: The AI Image Tool For People Who Actually Ship Designs
○Lesson 679Galileo: The UI Design Generator For Product Teams
○Lesson 680Uizard: The Napkin-Sketch-To-App Tool That Actually Works
○Lesson 681Runway: The AI Video Tool That Hollywood Actually Uses
○Lesson 682ElevenLabs: The AI Voice Platform That Redefined Audio
○Lesson 683Suno: The AI Music Tool That Made Everyone A Songwriter
○Lesson 684Descript: Edit Audio And Video By Editing The Transcript
○Lesson 685Pika: The AI Video Tool That Went Social-Native First
○Lesson 686Writer: The Enterprise Generative AI Platform For Content Teams
○Lesson 687Sudowrite: The AI Writing Tool Novelists Actually Love
○Lesson 688ShortlyAI: The Minimalist Writing Tool That Still Has Its Fans
○Lesson 689Zapier AI: When The Integration King Added Agents
○Lesson 690Motion: The AI Calendar That Rearranges Your Day Automatically
○Lesson 691Reclaim: The Calendar AI That's Calmer Than Motion
○Lesson 692Superhuman AI: The $30/Month Email App With AI Baked In
○Lesson 693ClickUp AI: The Everything-App That Added An Everything-AI
○Lesson 694Consensus: The AI Search Engine That Only Knows Science
○Lesson 695Elicit: The AI Research Assistant For Systematic Reviews
○Lesson 696Gong: The Revenue AI That Transformed Sales Teams
○Lesson 697Clay: The GTM Data Enrichment Tool That Changed Outbound
○Lesson 698Lindy: The No-Code Agent Platform For Business Automation
○Lesson 699Vic.ai: The AI That Does Your Accounts Payable
○Lesson 700Harvey: The AI Lawyers Actually Use
○Lesson 1207The Responses API: OpenAI's Modern Developer Surface
○Lesson 1209Structured Outputs: Make the Model Return Data You Can Trust
○Lesson 1210OpenAI Use-Case Playbook: Match the Surface to the Job
○Lesson 1345What Perplexity Is: Search-Augmented LLM, Not A Chatbot
○Lesson 1346Pro Search vs Default: When To Spend The Compute
○Lesson 1347Spaces: Building Team Knowledge Bases In Perplexity
○Lesson 1348Focus Modes: Academic, YouTube, Reddit, And When Each Wins
○Lesson 1349Citations And Source Verification: Perplexity's Biggest Win
○Lesson 1350Comet Browser: What It Does That Atlas And Operator Don't
○Lesson 1351Perplexity API: Building RAG Without Owning The Pipeline
○Lesson 1352Pages: Turning A Search Into A Sharable Doc
○Lesson 1353Perplexity For Journalism And Fact-Checking
○Lesson 1354Perplexity For Academic Research: Strengths And Limits
○Lesson 1355Switching The Underlying Model In Pro
○Lesson 1356Perplexity vs ChatGPT Search vs Google AI Overviews
○Lesson 1357Perplexity For Due Diligence On Companies And People
○Lesson 1358Daily-Brief Workflows In Perplexity
○Lesson 1359Perplexity For Travel Research: The Practical Playbook
○Lesson 1360Threads, Follow-ups, And Refining A Search
○Lesson 1361Sharing Perplexity Threads: Privacy And Accuracy
○Lesson 1362Perplexity Maker And Build Features
○Lesson 1363When Perplexity Hallucinates: Pattern-Spotting And Recovery
○Lesson 1364Building A Personal Research Stack With Perplexity At The Core
○Lesson 1365What Claude Code Is: Terminal-Native Agentic Coding
○Lesson 1366Installing And Authenticating Claude Code
○Lesson 1367The CLAUDE.md File: Project Persona And Rules
○Lesson 1368Slash Commands: Built-Ins And Custom
○Lesson 1369Subagents: When To Delegate vs Do It Yourself
○Lesson 1370Hooks: Automating Reactions To Tool Calls
○Lesson 1371Skills: Bundled Procedural Knowledge
○Lesson 1372MCP Servers: Adding New Capabilities
○Lesson 1373Settings.json: Permissions, Env Vars, Model Overrides
○Lesson 1374Plan Mode And ExitPlanMode
○Lesson 1375Background Tasks: Running Multiple Agents In Parallel
○Lesson 1376Worktrees: Isolated Agent Workspaces
○Lesson 1377Claude Code In CI And GitHub Actions
○Lesson 1378Claude Code IDE Integration: VS Code And JetBrains
○Lesson 1379The TodoWrite Tool: When It Actually Helps
○Lesson 1380Reading vs Editing: When To Use Read+Edit vs Write
○Lesson 1381Building A Custom Slash Command End-To-End
○Lesson 1382Claude Code For Code Review: The Security-Review Skill
○Lesson 1383Long-Context Strategies: When The Window Fills Up
○Lesson 1384Claude Code vs Codex vs Cursor vs Aider: The Honest Tradeoffs
○Lesson 1385Codex In 2026: OpenAI's Agentic Coding Layer
○Lesson 1386Codex CLI vs Codex Cloud: Picking The Right Surface
○Lesson 1387Setting Up Codex With Your Repo: AGENTS.md And Friends
○Lesson 1388Codex Review Mode: Pull-Request Review At Scale
○Lesson 1389Codex Tasks: Long-Running Asynchronous Work
○Lesson 1390Codex With Custom Tools And MCP
○Lesson 1391Understanding Codex Pricing — The Shape, Not The Sticker
○Lesson 1392Codex For Refactoring Legacy Code
○Lesson 1393Codex For Test Generation: From Coverage Gaps To Passing Suites
○Lesson 1394Codex For Framework Migrations: Pages To App, Vue 2 To 3, And Beyond
○Lesson 1395Codex Security Model: What Code It Can Run And Where
○Lesson 1396Codex vs Claude Code: Workflow Differences That Matter
○Lesson 1397Codex With Sandboxed Execution: Running Untrusted Code Safely
○Lesson 1398Multi-Repo Workflows In Codex
○Lesson 1399Codex For Technical Writing And Docs Generation
○Lesson 1400Codex For Incident-Response Triage
○Lesson 1401Codex Prompt Patterns That Actually Work
○Lesson 1402When Codex Fails: Debugging The Agent
○Lesson 1403Codex In A Regulated Environment
○Lesson 1404Building A Custom Codex Skill / Workflow
○Lesson 1612AI For Esports And Competitive Gaming
○Lesson 1700AI for Staying Connected With Family
○Lesson 1701AI for Medication Reminders You Will Actually Hear
○Lesson 1702AI for Travel Planning at Any Pace
○Lesson 1703AI for Hobbies: Gardening, Cooking, and Genealogy
○Lesson 1706Voice-First AI: Talking to a Computer Like a Person
○Lesson 1711AI for Hearing and Vision Help
○Lesson 1712AI for Staying Mentally Sharp
○Lesson 1722Group Chats With AI Assistants
○Lesson 1723Library and Community Resources for AI Learning
○Lesson 2510OpenClaw Heartbeats: Letting A Soul Think Without You
○Lesson 2511Time-Based And Event-Based Heartbeats: Choosing The Trigger
○Lesson 2512Heartbeat Budgets And Runaway Prevention
○Lesson 2513Debugging A Heartbeat Loop: Observability, Replay, And Failure Modes
○Lesson 2514Deploying OpenClaw: Local Box, Home Server, Or VPS
○Lesson 2515Observability: Logs, Traces, And Soul Timelines
○Lesson 2516Security: Sandboxing Skills, Least-Privilege Souls, Prompt-Injection Defense
○Lesson 2517Beyond The Basics: Federation, Custom Runtimes, Contributing Back
○Lesson 2518OpenClaw: Souls, Heartbeats, And Skills
○Lesson 2519Installing OpenClaw And Wiring It To A Local Model
○Lesson 2520Your First Soul: A Ten-Minute Hello World
○Lesson 2521OpenClaw Config And Project Layout
○Lesson 2522What A Skill Is In OpenClaw: Anatomy And Discovery
○Lesson 2523Building Your First OpenClaw Skill
○Lesson 2524Skill Registries, Sharing, And Trust
○Lesson 2525Composing Skills: When To Chain, When To Wrap, When NOT To
○Lesson 2526Designing A Soul: Voice, Values, And Constraints
○Lesson 2527Soul Memory Architecture: Episodic, Semantic, Procedural
○Lesson 2528Multi-Soul Orchestration: When To Split, How To Hand Off
○Lesson 2529Soul Evolution: When To Learn, Forget, Or Fork
○Lesson 4140Lovable Starts With A Product Brief
○Lesson 4150Cursor Rules: Teach The Editor Your Repo
○Lesson 4160Triangulate Sources With Perplexity
○Lesson 4161Comet And Browser Agent Safety
○Lesson 4170Claude Design For Fast Prototypes
○Lesson 4171Extract Design Tokens Before Screens Multiply
○Lesson 4172Run A Design Critique Loop
○Lesson 4173Accessibility Belongs In The Prototype
○Lesson 4174Handoff From Claude Design To Codex Or Claude Code
○Lesson 4180AGENTS.md Scope And Precedence In Codex
○Lesson 4182Delegate Background Work To Codex Cloud
○Lesson 4190Hermes As A Local Agent Brain
○Lesson 4200Your First OpenClaw Soul Should Be Boring
○Lesson 4210NanoClaw: Why Smaller Agent Runtimes Exist
○Lesson 4220Ollama Context Windows: Set Them Deliberately
○Lesson 22400Deploying Cursor at Team Scale: Adoption, Standards, and Cost Management
○Lesson 22401Claude Code Workflows: Beyond Single-Session Coding Help
○Lesson 22402Vercel AI Gateway: When Model Routing Beats Direct Provider Integration
○Lesson 22403LangGraph vs Custom Orchestration: When Frameworks Help and When They Hurt
○Lesson 22404LLM Observability Tools: What to Trace, What to Sample, What to Alert
○Lesson 25700Evaluating AI Tools for Your Stack: A Decision Framework
○Lesson 25701Deprecating AI Tools: How to Remove Things People Don't Use
○Lesson 25702BYOAI Policy: When Employees Use Their Own AI Tools
○Lesson 25703Tools for Defending Against Prompt Injection
○Lesson 25704AI Evaluation Platforms: When to Buy vs Build
○Lesson 27601RAG Framework Selection: LangChain, LlamaIndex, Custom
○Lesson 27602AI Agent Orchestration Frameworks Compared
○Lesson 27603AI Monitoring Stack: From Metrics to Quality
○Lesson 27604Eval Dataset Management: From Ad Hoc to Disciplined
○Lesson 28900AI Knowledge Base Platforms: Build, Buy, or Hybrid
○Lesson 28901AI Customer Support Platforms Compared
○Lesson 28902AI Dev Environment Tools: Cursor, Windsurf, Copilot
○Lesson 28903AI Ops Platforms: SRE in the AI Era
○Lesson 28904AI Marketing Platforms: Beyond ChatGPT for Content
○Lesson 29700AI Data Warehousing Tools: Snowflake AI, Databricks, BigQuery AI
○Lesson 29701No-Code AI Platforms: When They Fit
○Lesson 29702AI Gateway Services: Multi-Vendor Management
○Lesson 29703Prompt Management Platforms: Build vs Buy
○Lesson 29704LLM-as-Judge Platforms for Eval Automation
○Lesson 31100AI in Customer Data Platforms (CDP)
○Lesson 31101Marketing Automation With AI: Platform Selection
○Lesson 31102AI in Sales Engagement Platforms
○Lesson 31103AI in Recruitment Platforms: Bias and Compliance
○Lesson 31104AI in Design Platforms: Figma AI, Adobe Firefly
○Lesson 32600AI in Finance Platforms: Bloomberg, NetSuite, SAP
○Lesson 32601AI in Legal Platforms: Harvey, CoCounsel, Spellbook
○Lesson 32602AI in E-commerce Platforms: Shopify, BigCommerce, Salesforce Commerce
○Lesson 32603AI in Creative Platforms: Adobe Sensei, Figma AI
○Lesson 32604AI in Customer Service Platforms
○Lesson 34100AI in Cybersecurity Platforms
○Lesson 34101AI in DevSecOps Platforms
○Lesson 34102AI in Data Quality Platforms
○Lesson 34103AI in API Management Platforms
○Lesson 34104AI in Supply Chain Platforms
○Lesson 36000AI Coding Assistants in 2026: Cursor vs. Copilot vs. Claude Code vs. Windsurf
○Lesson 36001Comparing AI Evaluation Frameworks: Braintrust, Langfuse, Humanloop, Promptfoo
○Lesson 36002Vector Database Selection in 2026: Pinecone vs. Weaviate vs. pgvector vs. Turbopuffer
○Lesson 36003AI Gateway vs. Direct Provider APIs: When to Insert the Hop
○Lesson 36004AI Observability Stack 2026: Traces, Metrics, and Cost in One Pane
○Lesson 36006Autonomous Coding Agents 2026: Devin, Cline, OpenHands, and SWE-Bench Reality
○Lesson 36007AI Knowledge Base Platforms 2026: Glean vs. Notion AI vs. Custom RAG
○Lesson 36008AI Customer Support Platforms 2026: Intercom Fin, Decagon, Sierra, Ada
○Lesson 36009Writing an AI Tool Procurement Policy for a Growing Team
○Lesson 37500AI Incident Response Platforms for On-Call
○Lesson 37501AI Features in Product Analytics: Amplitude, Mixpanel, PostHog
○Lesson 37502AI in Spreadsheets: Excel Copilot, Google Sheets Gemini, Rows
○Lesson 37503AI Content Moderation: Hive, Perspective, OpenAI Moderation
○Lesson 37504AI Translation Platforms: DeepL, Google Translate, Lokalise AI
○Lesson 37505AI Meeting Summary Tools: Otter, Fireflies, Granola, Notion AI
○Lesson 37506AI Document Extraction: Reducto, Unstructured, Azure Document Intelligence
○Lesson 37507AI-Powered Developer Search: Sourcegraph Cody, Glean, Codeium Search
○Lesson 37508AI API Key Rotation and Secret Management Tools
○Lesson 37509AI Synthetic Data Platforms: Gretel, Mostly AI, Tonic
○Lesson 39000AI Feature Store Platforms: Tecton, Feast, Hopsworks
○Lesson 39001AI Model Serving Platforms: BentoML, Modal, Ray Serve, Replicate
○Lesson 39002AI Guardrails Platforms: Lakera, NeMo Guardrails, Guardrails AI
○Lesson 39004AI Fine-Tuning Platforms: OpenAI, Together, Fireworks, Anyscale
○Lesson 39005AI Tracing Platforms: Langfuse, LangSmith, Helicone, Phoenix
○Lesson 39006AI Dataset Versioning Platforms: DVC, LakeFS, Pachyderm
○Lesson 39007AI Secret Scanning Platforms: GitGuardian, TruffleHog, Doppler Scan
○Lesson 39008AI Vector Index Management: Pinecone, Weaviate, Qdrant, pgvector
○Lesson 39009AI LLM Routing Platforms: Martian, Not Diamond, OpenRouter
○Lesson 40500AI Agent Evaluation Platforms in 2026
○Lesson 40501AI Agent Runtime Platforms in 2026
○Lesson 40502AI Batch Inference Platforms for Bulk Workloads
○Lesson 40503AI Code Review Bot Platforms in 2026
○Lesson 40504Comparing Embeddings Providers Beyond OpenAI
○Lesson 40505Enterprise LLM Gateways: Portkey, LiteLLM, Vercel AI Gateway
○Lesson 40506On-Prem Inference Platforms for Regulated Industries
○Lesson 40507AI Prompt Testing Platforms vs Rolling Your Own
○Lesson 40508Comparing Hosted RAG Platforms in 2026
○Lesson 40509Voice Agent Platforms: Vapi, Retell, Bland in 2026
○Lesson 42400Comparing edge AI deployment platforms (Cloudflare, Fastly, Vercel)
○Lesson 42401Evaluating prompt injection scanners for production AI apps
○Lesson 42402Comparing managed RAG platforms (Pinecone, Vectara, Mongo Atlas)
○Lesson 42404Using feature flag platforms (LaunchDarkly, Statsig) for AI rollouts
○Lesson 42405Choosing a secrets vault for AI agent credentials
○Lesson 42407Allocating AI costs across teams with platforms like Vantage and CloudZero
○Lesson 43904AI Fine-Tuning Platforms: OpenAI vs Together vs Databricks vs DIY
○Lesson 43905AI Multi-Modal Platforms: Image, Audio, Video Toolchains
○Lesson 43906AI Coding Agent Platforms: Cursor, Cline, Aider, Devin
○Lesson 43907AI Data Labeling Platforms: Scale, Surge, Snorkel, Label Studio
○Lesson 43908AI On-Device Inference: Core ML, ONNX Runtime, MLC LLM
○Lesson 43909AI Agent Memory Platforms: Mem0, Zep, Letta
○Lesson 44300AI feedback collection platforms
○Lesson 44301AI canary testing platforms
○Lesson 44302AI data labeling platforms
○Lesson 44303AI experiment tracking platforms
○Lesson 44304AI rate limit management tools
○Lesson 44305AI shadow deployment tools
○Lesson 44306AI cost attribution tools
○Lesson 44307AI context management platforms
○Lesson 44308AI tool call debugging tools
○Lesson 44309AI output watermarking tools
○Lesson 45822AI Guardrail Libraries: NeMo Guardrails, Guardrails AI, Lakera
○Lesson 45823AI RAG Frameworks: LlamaIndex, Haystack, and Building Your Own
○Lesson 45824AI Agent Orchestration: LangGraph, CrewAI, and AutoGen Compared
○Lesson 45825AI Model Routers: OpenRouter, Portkey, and the AI Gateway Pattern
○Lesson 45826AI Document Extraction: Reducto, Unstructured, and the OCR Stack
○Lesson 45828AI Browser Agents: Browserbase, Browserless, and Stagehand
○Lesson 45829AI Red-Team Platforms: HiddenLayer, Robust Intelligence, Lakera Red
○Lesson 46300AI tools: how to choose an AI coding assistant for your team
○Lesson 46301AI tools: pair-programming workflows that don't slow you down
○Lesson 46302AI tools: RAG vs fine-tuning — picking the right adaptation
○Lesson 46303AI tools: vector databases without the hype
○Lesson 46305AI tools: cost-control patterns for LLM features
○Lesson 46306AI tools: evaluation platforms and what to look for
○Lesson 46308AI tools: running local models and when it pays off
Lesson 46309AI tools: MCP and the rise of standard tool protocols
○Lesson 47802Cursor Background Agents: Letting AI Code While You Sleep
○Lesson 47803Lovable App Builder: When AI Spec-to-App Is Enough
○Lesson 47804Modal: Serverless GPUs for AI Without Kubernetes
○Lesson 47806Replicate: Hosting Open AI Models Without Owning GPUs
○Lesson 47807Perplexity Pro: AI Research Search With Sources You Can Verify
○Lesson 47808ElevenLabs Voice Cloning: Production Voiceover With Consent Discipline
○Lesson 47809Anthropic Batch API: Half-Price Claude for Async Workloads
○Lesson 48300AI Tools: Pick the Right IDE AI Mode for the Work In Front of You
○Lesson 48301AI Tools: Use Context Files (.cursorrules, AGENTS.md, CLAUDE.md) Without Bloat
○Lesson 48303AI Tools: Evaluate a New Coding Agent Without Marketing Bias
○Lesson 48304AI Tools: When to Reach for a CLI Coder vs an IDE vs a Web App
○Lesson 48305AI Tools: Keep Secrets Out of Prompts, Logs, and Vendor Telemetry
○Lesson 48306AI Tools: Track Cost Per Developer Per Month and Justify the Spend
○Lesson 48307AI Tools: Decide Between Local Models and Hosted APIs With a Real Workload
○Lesson 48308AI Tools: Pick an Eval Platform You Will Actually Use
○Lesson 48309AI Tools: Reduce AI Vendor Lock-In Without Adding Useless Abstraction
○Lesson 49800Anthropic Claude Skills: Packaging Domain Procedures the Model Can Pick Up
○Lesson 49801OpenAI Responses API for Reasoning Models: Carrying State Across Turns
○Lesson 49802Google Vertex Model Garden: Picking Among First-Party and Open Models
○Lesson 49803Azure AI Foundry Evaluations: Promotion-Gates for Enterprise Models
○Lesson 49804Anthropic Message Batches API: Spending Half-Price on Patient Workloads
○Lesson 49805OpenAI Realtime API for Voice Agents: Streaming Speech Both Ways
○Lesson 49806LangGraph for Stateful Agents: Modeling Loops, Forks, and Checkpoints
○Lesson 49807Weights and Biases Weave: Tracing AI Apps Across Calls and Versions
○Lesson 49809LM Studio and Ollama for Local Models: Running AI on the Desktop Honestly
○Lesson 50300AI and choosing an IDE assistant
○Lesson 50301AI and using the CLI coding tools
○Lesson 50304AI and prompt management platforms
○Lesson 50305AI and evaluation frameworks
○Lesson 50306AI and image generation tool comparison
○Lesson 50307AI and video generation workflow pick
○Lesson 50308AI and voice cloning tools with consent
○Lesson 50309AI and self-hosted LLM deployment tools
○Lesson 51800AI Tool Langfuse for Prompt Management: Versioning Prompts in Production
○Lesson 51801AI Tool vLLM Serving Configuration: Tuning for Real Traffic
○Lesson 51802AI Tool pgvector RAG Pipeline: Drafting an Indexing and Query Plan
○Lesson 51803AI Tool LlamaIndex Router Query Engine: Picking the Right Tool
○Lesson 51804AI Tool Haystack Pipeline Evaluation: Measuring End-to-End Quality
○Lesson 51805AI Tool Promptfoo Config Suite: Running Side-by-Side Prompt Tests
○Lesson 51806AI Tool Temporal for Agent Workflows: Drafting Durable Loops
○Lesson 51807AI Tool Modal for Distributed Evaluation: Drafting a Fan-Out Job
○Lesson 51808AI Tool Weaviate Hybrid Search: Combining Keyword and Vector Recall
○Lesson 51809AI Tool OpenLLMetry Tracing Setup: Instrumenting LLM Calls End to End
○Lesson 53800AI Tools: vLLM Prefix Caching for Throughput
○Lesson 53801AI Tools: TensorRT-LLM Quantization Pipelines
○Lesson 53802AI Tools: Ray Serve LLM Multiplexing
○Lesson 53803AI Tools: Langfuse Trace-Linked Evals
○Lesson 53804AI Tools: MLflow 3 GenAI Prompt Registry
○Lesson 53805AI Tools: BentoML Quantized Deployment
○Lesson 53806AI Tools: pgvector Half-Precision Indexes
○Lesson 53807AI Tools: Instructor for Structured Outputs
○Lesson 53808AI Tools: Promptfoo Red-Team Test Suites
○Lesson 53809AI Tools: DSPy Program Compilation
○Lesson 55800AI and Cursor Rules .mdc Tuning for Team Repos
○Lesson 55802AI and Codex CLI Pipeline Integration
○Lesson 55803AI and Perplexity Research Mode Discipline
○Lesson 55804AI and Lovable Component Export Tuning
○Lesson 55805AI and Ollama Local Model Routing for Mixed Workloads
○Lesson 55806AI and Claude Design Component Token Mapping
○Lesson 55807AI and Hermes Message Routing Policy for Agents
○Lesson 55808AI and OpenClaw Skill Bundling for Team Reuse
○Lesson 55809AI and Vercel Cron Observability for Scheduled AI Jobs
○Lesson 56301Picking a Vector Store for Your Scale
○Lesson 56302Building a Lightweight Eval Harness
○Lesson 56303Tracing Every LLM Call With Inputs and Costs
○Lesson 56304When Fine-Tuning Beats Prompting (and When It Doesn't)
○Lesson 56306Using Prompt Caching to Cut Cost and Latency
○Lesson 56307Designing Streaming UX That Survives Model Errors
○Lesson 56308Handling Provider Rate Limits Without Hurting Users
○Lesson 56309Keeping Secrets Out of Prompts and Logs
○Lesson 58300AI Canvas vs Chat Mode: When to Switch Interfaces
○Lesson 58301AI Vision for Document Extraction: PDFs to Structured Data
○Lesson 58302AI Voice Mode for Meeting Prep and Debriefs
○Lesson 58303AI Tab Completion: Cursor, Copilot, and Inline Suggestions
○Lesson 58304AI Image Editing vs Generation: Two Different Workflows
○Lesson 58305Deep Research Modes: When to Wait 10 Minutes for an AI Report
○Lesson 58306AI Projects and Custom Memory: Persistent Context Across Chats
○Lesson 58307AI Agent Mode vs Chat: When to Hand Over the Wheel
○Lesson 58309AI for Spreadsheet Formulas: From Description to FORMULA
○Lesson 58310AI Video Summarization: From Hour-Long Recordings to Notes
○Lesson 58312AI Batch Processing: Run 1,000 Prompts Cheaply
○Lesson 58313AI Evals: Testing AI Outputs Like You'd Test Code
○Lesson 58314Fine-Tune vs Prompt: When AI Tuning Pays Off
○Lesson 58315AI Model Routers: Pick the Right Model Per Task
○Lesson 58316AI Prompt Caching: 90% Discount on Repeated Context
○Lesson 58317AI Streaming vs Block Responses: UX Tradeoffs
○Lesson 58318AI Tool Use: Letting the Model Call Functions
○Lesson 58319AI Screenshot-to-Code: From Mockup to Component
○Lesson 58320Local AI Models: When to Run Llama or Mistral on Your Laptop
○Lesson 58321AI Image Style References: Lock Visual Identity Across Generations
○Lesson 58322AI Realtime APIs: Voice-In, Voice-Out at Conversation Speed
○Lesson 58323AI Browser Automation: Operator, Computer Use, and Browser Agents
○Lesson 58324AI Content Detectors: Why You Shouldn't Trust Them
○Lesson 60600AI Tool: Cursor for Codebase-Aware Editing, Part 1

Curriculum
·
Creators
·
Tools Literacy
·
AI tools: MCP and the rise of standard tool protocols

Lesson 1709 of 2116

AI tools: MCP and the rise of standard tool protocols

Standard protocols like MCP let one agent talk to many tools without bespoke glue. Adopt them when your tool count grows past a handful.

CreatorsTools Literacy~24 min readBI2 · Representation & ReasoningBI3 · LearningBI4 · Natural InteractionPrint / PDF

Big idea

Standard protocols like MCP let one agent talk to many tools without bespoke glue. Adopt them when your tool count grows past a handful.

Lesson map

What this lesson covers

40 min45 blocks10 concepts

Learning path

The main moves in order

1The premise
2AI Tools: Add an MCP Server Only When the Tool Earns Its Slot
3The premise
4Choosing Which MCP Servers to Expose to Your Agent

Concept cluster

Terms to connect while reading

MCPtool protocolsinteroperabilitytool inflationcontext costscoped servers

Read4

Sections15

Lists8

Notes16

Terms2

Section 1

The premise

When each tool integration is hand-coded, every new tool is a project. Standard protocols (Model Context Protocol and similar) let agents discover and call tools through a uniform surface, dramatically lowering integration cost.

What AI does well here

Discover tools advertised by an MCP-compliant server
Call tools using the protocol's schema
Parse tool responses in the protocol's format

Try this adoption signal

Adopt MCP when: (1) you're integrating >5 tools across teams, (2) you want third-party tool ecosystems, (3) you're tired of bespoke per-tool clients. Skip it if your agent only needs 2-3 stable tools.

Check-in 1. Got it so far?

What AI cannot do

Make a non-MCP tool MCP-compliant on its own
Guarantee that every MCP server is well-built or safe
Replace authentication and authorization design

Watch out: untrusted MCP servers

Connecting your agent to a third-party MCP server hands it a new attack surface. Vet servers like you'd vet any third-party dependency, and limit which servers each agent can reach.

Key terms in this lesson

MCP
tool protocols
interoperability

Check-in 2. Got it so far?

Evaluate systematically

Before adopting any AI tool: check the data policy, benchmark on your actual use cases, and plan an exit strategy. Vendor lock-in with AI tools can be painful.

Lesson complete

You've completed "AI tools: MCP and the rise of standard tool protocols". Mark this lesson done and keep going — every lesson builds on the last.

Section 2

AI Tools: Add an MCP Server Only When the Tool Earns Its Slot

Section 3

The premise

MCP makes connecting tools easy, which is exactly why teams over-add them; every additional server costs context tokens and increases the chance the agent picks the wrong tool.

Check-in 3. Got it so far?

What AI does well here

Estimate the context cost of a new MCP server
Decide if the value justifies tool-choice complexity
Recommend disabling unused servers per project
Audit installed servers monthly

Prompt: server triage

List MCP servers you have installed and how often you use each. Ask: 'Rank by value-per-context-token. Recommend which to disable, which to scope to specific projects, and what is missing.'

What AI cannot do

Make the model use the right server every time
Replace permission scoping on the underlying systems
Test compatibility across IDE versions for you

Check-in 4. Got it so far?

Disable, do not just ignore

An unused MCP server still loads tool definitions into context. Disable per project so you only pay for what you use today.

Evaluate systematically

Before adopting any AI tool: check the data policy, benchmark on your actual use cases, and plan an exit strategy. Vendor lock-in with AI tools can be painful.

Lesson complete

You've completed "AI Tools: Add an MCP Server Only When the Tool Earns Its Slot". Mark this lesson done and keep going — every lesson builds on the last.

Check-in 5. Got it so far?

Section 4

Choosing Which MCP Servers to Expose to Your Agent

Section 5

The premise

Exposing every available MCP server bloats context and confuses tool selection. Pick the smallest set that covers the agent's job.

What AI does well here

Use exposed MCP tools when they clearly match the task.
Refuse a task when no exposed tool fits.

MCP shortlist process

For each agent, list the 5-10 user jobs. For each job, pick at most 1-2 MCP tools. Disable everything else for that agent.

Check-in 6. Got it so far?

What AI cannot do

Self-discover and add new MCP servers safely.
Pick well from dozens of overlapping tools.

Audit the permissions each MCP grants

An MCP server that 'just reads' may also write. Read its scopes before you connect it to a production agent.

Evaluate systematically

Before adopting any AI tool: check the data policy, benchmark on your actual use cases, and plan an exit strategy. Vendor lock-in with AI tools can be painful.

Check-in 7. Got it so far?

Lesson complete

You've completed "Choosing Which MCP Servers to Expose to Your Agent". Mark this lesson done and keep going — every lesson builds on the last.

Section 6

AI MCP Connectors: Plug AI into Your Existing Tools

Section 7

The premise

MCP-style connectors let one AI client (Claude Desktop, Cursor, etc.) talk to many apps with a single auth dance per tool.

What AI does well here

Read content from connected tools (Drive, Notion, Slack).
Take simple actions (post, draft, update) when authorized.
Cite the source tool and document.
Chain across tools in one prompt (read Notion, post Slack).

Check-in 8. Got it so far?

MCP-aware prompt

Say: 'Using my connected Notion + Slack: find the latest project doc in Notion under [page] and post the TL;DR to Slack channel #[name]. Show me what you'll post before sending.'

What AI cannot do

Replace deep app-native features (formulas, automations).
Survive auth-token expiry without re-auth.

Connectors expand attack surface

Each MCP connector is an open door into your tools. Audit which connectors you've granted, and revoke unused ones.

Check-in 9. Got it so far?

Evaluate systematically

Before adopting any AI tool: check the data policy, benchmark on your actual use cases, and plan an exit strategy. Vendor lock-in with AI tools can be painful.

Lesson complete

You've completed "AI MCP Connectors: Plug AI into Your Existing Tools". Mark this lesson done and keep going — every lesson builds on the last.

Key terms in this lesson

MCP
tool protocols
interoperability
tool inflation
context cost
scoped servers
tool-surface
scope
connectors
integration

End-of-lesson quiz

Check what stuck

15 questions · Score saves to your progress.

Tutor

Curious about “AI tools: MCP and the rise of standard tool protocols”?

Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.

Your question

Try one:

Progress saved locally in this browser. Sign in to sync across devices.

Related lessons

Keep going

Creators · 10 min
Codex With Custom Tools And MCP
Codex's real power shows when you connect it to your own tools — internal APIs, datastores, ticketing systems — usually via Model Context Protocol.
Creators · 10 min
Deploying Cursor at Team Scale: Adoption, Standards, and Cost Management
Individual Cursor adoption is easy; team deployment requires shared standards (rules files, MCP servers), security review, and cost management at scale.
Creators · 11 min
AI in DevSecOps Platforms
DevSecOps platforms integrate security into deployment. AI accelerates while maintaining security gates.

Previous: AI tools: running local models and when it pays off

Next module: Next

Report an error

Reading mode