CODING TOOLS (Modules 1–9)

All prices verified June 2026. ⚠ verify = re-check before reselling.

MODULE 1: Cursor CATEGORY: Coding DEPTH: CORE ONE-LINE SUMMARY: A VS Code fork with an AI agent and a best-in-class autocomplete ("Tab") model baked into the editor. VS CLAUDE CODE: Same agentic editing power, but inside a visual IDE with inline diffs, a side-chat, and predictive Tab autocomplete as you type. Claude Code is terminal-first and headless; Cursor is GUI-first. Cursor's "Tab" (multi-line, next-edit prediction) is the thing Claude Code has no equivalent of. BEST FOR:

• Daily driving a large TS/React/Next codebase with mouse + keyboard, not just terminal.
• Fast inline edits where you want to see the diff in the file before accepting.
• Devs who want the agent + autocomplete + GUI in one window (the "IDE-native CC"). WEAKNESS: Credit/usage anxiety — the $20 Pro plan is now metered as $20 of model credits, so heavy agent use burns out mid-month and you hit slow/queued requests. Forking VS Code means occasional lag behind upstream VS Code features. COST: ⚠ verify — Free (Hobby) tier; Pro $20/mo ($16/mo annual) with $20 of monthly model credits + unlimited Tab; Business higher. Power users routinely exceed the credit pool. HANDS-ON TASK: Install Cursor, open one of your existing Next.js repos, press Cmd/Ctrl+K on a function and ask it to refactor; then accept a few Tab completions while typing a new component. Note how the Tab model predicts your next edit location, not just the next token. GOTCHA: The $20 you pay is also your model spend. A single long agent session on a big repo can eat days of credits — budget it or you'll be rate-limited when you need it most.

MODULE 2: Windsurf CATEGORY: Coding DEPTH: SKIM ONE-LINE SUMMARY: Cursor's main rival — a VS Code-based AI editor whose pitch is a smoother, more autonomous agent ("Cascade") that keeps shared context across your whole flow. VS CLAUDE CODE: Like Cursor — GUI IDE with an agent — but the agent (Cascade) leans more "do it all for me" and the autocomplete is unlimited on every tier. Claude Code gives you more granular step approval; Windsurf optimizes for flow/autonomy. BEST FOR:

• Devs who found Cursor's UX busy and want a calmer, more guided agent.
• Unlimited tab autocomplete without watching a credit meter.
• Teams wanting a quota model (daily/weekly) instead of credit math. WEAKNESS: Smaller ecosystem/mindshare than Cursor; went through a disruptive March 2026 pricing overhaul (credits → quotas) and a corporate ownership saga that spooked some users. COST: ⚠ verify — Free tier (light daily/weekly quota, unlimited Tab); Pro $20/mo (~$12/mo annual); Max $200/mo; add-on credits $10/250 units. (Switched from credits to daily/weekly quotas in March 2026.) HANDS-ON TASK: Open the same repo you used in Cursor, give Cascade a multi-file task ("add a loading skeleton to every page in /app"), and compare how much it does before pausing vs. Cursor. GOTCHA: Pricing model changed under users mid-2026. If you teach or sell Windsurf, the plan names/quotas you learned may be stale within a quarter — always link the live pricing page.

MODULE 3: Bolt.new CATEGORY: Coding DEPTH: SKIM ONE-LINE SUMMARY: Prompt-to-app in the browser — describe an app, it scaffolds a full-stack (often Vite/React/Node) project in a live in-browser environment and can deploy it. VS CLAUDE CODE: Claude Code edits your local repo; Bolt runs the entire dev environment in the browser (via WebContainers) and handles install/run/deploy for you. No local setup, but you trade away surgical control over a real large repo. Greenfield-first. BEST FOR:

• Zero-setup throwaway prototypes and demos you can share via URL in minutes.
• Validating a micro-SaaS idea before you commit a real local project.
• Letting the tool own the whole build→preview→deploy loop while you stay in the prompt. WEAKNESS: Token burn is brutal on iteration — each fix re-sends context and eats tokens fast; it struggles as the project grows; debugging deep issues is painful vs. a real editor. COST: ⚠ verify — Free (1M tokens/mo, daily cap ~150K); Pro $25/mo (~13M tokens, raised from 10M in May 2026); Teams $30/member/mo. Token rollover + custom domains on paid. HANDS-ON TASK: Build a single-page "waitlist + email capture" landing app from one prompt, deploy it, and copy the result into a local repo — feel where the in-browser model helps and where it fights you. GOTCHA: Token-metered iteration means fixing bugs costs money. A stubborn bug can drain your monthly tokens in an afternoon. Get the prompt right up front; don't ping-pong.

MODULE 4: Lovable CATEGORY: Coding DEPTH: SKIM ONE-LINE SUMMARY: Prompt-to-app aimed at non-/semi-technical builders — generates a working full-stack app (React + Supabase) with a friendly chat UI and visual editing. VS CLAUDE CODE: Claude Code assumes a developer; Lovable assumes you might not be one. It hides the stack behind a chat + visual editor and wires up DB/auth (Supabase) for you. You lose low-level control; you gain a client-friendly handoff. BEST FOR:

• Building an app a non-technical client can keep poking at after you deliver.
• Fast CRUD SaaS MVPs where Supabase auth/DB is the default you want anyway.
• Selling "I'll build your app" to clients who want to see it grow visually. WEAKNESS: Credit-metered and the real monthly cost balloons because Cloud (DB, bandwidth) and in-app AI inference are billed separately on usage. Less control once the app gets complex; export/escape-hatch friction. COST: ⚠ verify — Free (150 credits/mo cap); Pro $25/mo (100–150 credits); Business $50/mo (SSO, security center). Cloud + AI usage billed on top. HANDS-ON TASK: Generate a "personal CRM" app, connect the Supabase backend it offers, add one auth-gated page — then check the credit meter to feel the real cost curve. GOTCHA: The sticker price is not the bill. A live production Lovable app pays separately for storage, bandwidth, and model inference. Quote clients on total cost, not the plan price.

MODULE 5: v0 (by Vercel) CATEGORY: Coding DEPTH: CORE (you're on Vercel/React — this is your native fit) ONE-LINE SUMMARY: A generative UI tool — describe a component or paste a Figma frame, get production-ready React + Tailwind + shadcn/ui you paste straight into your Next app. VS CLAUDE CODE: v0 is narrow on purpose: it does design→React UI better than a general agent. Claude Code can write a component, but v0 gives you polished, on-brand, shadcn-styled UI with visual iteration and Figma import. Use v0 for the front of the house, Claude Code for the rest. BEST FOR:

• Generating dashboard/landing/form UI in your exact stack (Next + Tailwind + shadcn) and pasting it in.
• Turning a Figma frame into real components without hand-translating.
• Rapid UI exploration before wiring logic in Claude Code. WEAKNESS: It's a UI generator, not an app builder — weak on backend/business logic and large-codebase awareness. Credit model means heavy generation costs add up. COST: ⚠ verify — Free ($5/mo credits); Premium $20/mo ($20 credits, Figma import, v0 API); Team $30/user/mo. Mini/Pro/Max model tiers consume credits differently. HANDS-ON TASK: Generate a pricing-page section in v0, copy the JSX into one of your Next projects, and confirm the shadcn/Tailwind classes drop in clean. Time how much faster than hand-building. GOTCHA: v0 output assumes shadcn/ui + Tailwind conventions. Paste it into a project with a different design system and you'll spend the saved time un-theming it. Standardize on shadcn to make v0 free money.

MODULE 6: Replit Agent CATEGORY: Coding DEPTH: REFERENCE ONE-LINE SUMMARY: An agent inside Replit's cloud IDE that builds, runs, debugs, and hosts an app — with DB, deployment, and collaboration all in one browser tab. VS CLAUDE CODE: Replit is build + run + host + DB in the cloud; Claude Code is just the coding. If you want "from prompt to live URL with a database, no local anything," Replit does the whole platform. You give up local control and pay per credit. BEST FOR:

• Teaching/demo or hackathon builds where hosting + DB in one place wins.
• Mobile or no-local-setup contexts (you can build from a tablet).
• Quick internal tools that can live on Replit's hosting. WEAKNESS: Credits expire monthly (no rollover on Core); cost is unpredictable on heavy agent use; not where you'd run a serious production app long-term. COST: ⚠ verify — Free (daily Agent credits, 1 published project); Core $20/mo ($25 credits, no rollover); Pro ~$95/mo annual ($100 credits, rollover 1mo). COMMUNICATION SHORTCUT: "Replit Agent is the all-in-one cloud builder — agent + IDE + database + hosting in one tab. Great for zero-setup and teaching, but it's a platform you rent, not a repo you own." GOTCHA: Credits vanish at cycle end on Core. Don't bank them; you can't.

MODULE 7: GitHub Copilot CATEGORY: Coding DEPTH: SKIM ONE-LINE SUMMARY: The original AI pair-programmer — inline autocomplete and chat across VS Code/JetBrains, now with agent mode, deeply integrated with GitHub. VS CLAUDE CODE: Copilot is completion-first (free, unlimited inline suggestions) with agent features bolted on; Claude Code is agent-first. Copilot is the cheap always-on autocomplete layer; Claude Code is the heavy lifter. Many devs run both. BEST FOR:

• Cheap, unlimited inline autocomplete in any editor while you hand-drive.
• Teams already living in GitHub (PR summaries, code review, Copilot CLI).
• A low-cost baseline AI in editors where you don't run a full agent. WEAKNESS: Agent mode is weaker/less autonomous than Claude Code or Cursor. June 2026 move to usage-based billing (AI Credits) made heavy chat/agent use a metered cost and annoyed users. COST: ⚠ verify — Free tier; Pro $10/mo ($10 AI Credits); Pro+ $39/mo; Business $19/user/mo; Enterprise $39/user/mo. Inline completions are FREE and don't burn credits; only chat/agent/review/CLI do (since June 1, 2026). HANDS-ON TASK: Enable Copilot in VS Code alongside Cline (Module 8). Use Copilot only for inline completions and Cline for agentic tasks — feel the natural division of labor. GOTCHA: Post-June-2026, "Copilot" means two cost models in one product: free completions + metered agent/chat. People conflate them. Be precise when advising clients.

MODULE 8: Cline (VS Code extension) CATEGORY: Coding DEPTH: CORE (closest VS Code analog to your Claude Code workflow) ONE-LINE SUMMARY: An open-source VS Code extension that runs a transparent agentic loop — plans, edits files, runs terminal commands, asks approval — using your own API key (Anthropic, OpenAI, etc.). VS CLAUDE CODE: Cline is basically Claude Code's agentic loop inside VS Code's GUI, BYO key. You see every planned step and approve it (like CC's permission flow) but with diffs in the editor. No subscription — you pay raw model API costs. If you love CC but want it in the IDE without Cursor's fork, Cline is it. BEST FOR:

• Bringing your CC-style agent workflow into VS Code with full step visibility.
• BYOK cost control — pay exact API usage, no platform markup, swap models freely.
• Auditable agent runs where you want to see/approve each file edit and command. WEAKNESS: You manage your own API keys and spend (no flat cap — a runaway loop = a real bill). Less polished than Cursor; setup is more hands-on. COST: Extension is free/OSS. You pay your model provider directly (e.g., Anthropic/OpenAI API). Cost = your token usage; no subscription. HANDS-ON TASK: Install Cline, plug in an Anthropic API key, point it at a repo, and give it the same multi-file task you gave Cursor's agent. Compare the approval/diff UX and the actual $ cost. GOTCHA: BYOK means no spending ceiling. A big agentic task on a large repo can quietly run up API cost. Set provider-side budget alerts. (You already think this way from Claude Code.)

MODULE 9: Aider CATEGORY: Coding DEPTH: SKIM ONE-LINE SUMMARY: An open-source terminal AI pair-programmer that edits your local files and auto-commits each change to git — scriptable, model-agnostic, BYOK. VS CLAUDE CODE: The closest philosophical cousin — terminal, real files, agentic edits. Differences: Aider is OSS/free, leans on tight git integration (every edit = a commit), and is highly scriptable/pipeable for batch jobs. Less polished UX, more hacker tool. Think "CC, but free, git-obsessed, and scriptable." BEST FOR:

• Cheap, scriptable batch edits across a repo (BYO cheap model like DeepSeek).
• Git-disciplined workflows where you want every AI change as a discrete commit.
• CLI automation / CI pipelines that need an AI editor as a step. WEAKNESS: Bare-bones UX; no GUI; less capable at sprawling multi-file reasoning than CC/Cursor; you wire up your own model and keys. COST: Free/OSS. Pay your model API directly. Pairs well with DeepSeek for near-zero-cost edits. HANDS-ON TASK: pip install aider-chat, point it at a small repo with a cheap model, ask for a refactor, and watch it auto-commit. Try piping a task in non-interactively. GOTCHA: Auto-commit-everything is great until it isn't — on a dirty/uncommitted repo it can entangle your changes with its commits. Start from a clean tree.

Coding tools — at a glance

Sources (verified June 2026)

• Cursor: cloudzero, nocode.mba
• Windsurf: cloudzero, verdent
• Bolt/Lovable: nocode.mba bolt, nocode.mba lovable
• v0: uibakery
• Replit: nocode.mba
• Copilot: GitHub blog, plans

Next: 02-llms.md.