Shape components yourself. Hand the sweep to the agent.

Frontend work rewards a different tool balance than backend work: much of it is visual and iterative, which is exactly where an editor-based agent with Tab completions and inline edit earns its keep. The recipes below cover React, Vue, Angular, Next.js and Nuxt, plus the mobile-web patterns that sit alongside them.

$59.99 once for 12 months of access. Nothing renews on its own. 30-day refund on a first purchase.

Pages picked for this work
6
Steps on the reading path
16
Questions in the free scorecard
25
By-role data last updated
2026-08-17

Pages: React recipes, Next.js recipes, Vue recipes, Nuxt recipes, Angular recipes, Mobile-web patterns.

One change, two ways to work.

One frontend sweep, drawn two ways: a prop renamed across forty components.

  • Written by you
  • Read by you
  • Agent and checks

With Tab and inline edit

Forty edits, one at a time.

Tab completions and Cmd+K inline edit suit a component you are shaping, because you see the change at once. Across forty components they are forty separate edits.

  1. Component You edit
  2. Component You edit
  3. Component You edit
  4. … 40 in all You edit

As one agent task

One sweep, one loop.

The sweep is agent work: all forty components in one task, with the checks the recipe names run until they pass. You read the evidence it produces, and the code only for an escalation class.

Prompt

One instruction

Each recipe pairs a prompt with the gate that proves the result.

You write and read

One task for the agent

  • Component
  • Component
  • Component
  • … 40 in all

Runs the type check, lint and tests until they pass.

Agent writes, checks read

Evidence

Gate results and the report

The type check, the tests and the report. A change in an escalation class, such as authentication or money, still gets a named reader of the code.

You read

Iteration speed matters more than autonomy. A component you are shaping benefits from Tab completions and Cmd+K inline edit — a change you can see immediately — more than from an agent that disappears for two minutes and returns a diff. The exception is the sweep: a design-token rename, a prop renamed across forty components, a framework migration. Those are agent work: one task, and gate results and a report to read, instead of the same edit forty times.

Drawing
One change, two ways to work.
Source
Drawn from the by-role text for frontend developers, last updated 2026-08-17; the gates are the ones the React recipe names.
Note
An illustration of the by-role data, not a measurement

The failures your tests do not catch.

Each recipe opens with the failure it exists to prevent. Here are 5, in the recipes’ own words, with the page that fixes them.

  1. The scene
    “You ask the agent for a data table and it writes useReactTable, which TanStack Table v9 no longer exports. You ask for a form and get z.string().email(), deprecated in Zod 4.”

    The page that fixes it React recipesCopy-paste prompts for React 19 work in Claude Code, Codex and Cursor: Figma to code, TanStack v9 tables, forms, optimistic UI, and gates.

    Open React recipes 2026-10-12

  2. The scene
    “The agent’s dashboard renders, but it creates middleware.ts with export const runtime = 'edge', calls revalidateTag('posts') with one argument, and puts 'use client' at the top of a page. Everything compiles, the build warns, and nobody reads the warnings.”

    The page that fixes it Next.js recipesCopy-paste prompts that get Next.js 16 App Router code from Claude Code, Codex or Cursor, with the build, type and Playwright gates that prove it.

    Open Next.js recipes 2026-09-26

  3. The scene
    “You ask the agent for a store and it writes Vuex with the Options API. You ask for a form and it installs @vee-validate/zod, which declares Zod 3 as its peer while your project runs Zod 4.”

    The page that fixes it Vue recipesCopy-paste prompts for Vue 3.5 work in Claude Code, Codex and Cursor: Pinia 4 stores, composables, forms, routes, i18n, and the gates that prove them.

    Open Vue recipes 2026-10-12

  4. The scene
    “The agent’s feature works on your machine, but it put the pages in pages/ instead of app/pages/, passed a Zod schema straight to readValidatedBody, and read document.cookie in a plugin that also runs on the server.”

    The page that fixes it Nuxt recipesCopy-paste prompts that get Nuxt 4 code from Claude Code, Codex or Cursor, with the Nuxt docs MCP server and the type, test and build gates that prove it.

    Open Nuxt recipes 2026-10-12

  5. The scene
    “… the agent produces a component that compiles, the spec passes, and the view never updates, because the agent mutated a plain field the way Zone.js-era Angular allowed, and your app has no Zone.js to notice.”

    The page that fixes it Angular recipesCopy-paste prompts that get Angular 22 code from Claude Code, Codex or Cursor: signals, Signal Forms, zoneless, SSR, proven by build, Vitest and E2E gates.

    Open Angular recipes 2026-10-12

Scenes are the opening scenarios of the pages and are written as examples, not as reader reports. Quotations are trimmed. The pages are documentation pages and come with the Toolkit.

Your reading path, in the order it pays off.

From the 16 steps on the developer path, 5 picked for work you see in the browser: a plan you read before a sweep, and checks the agent cannot weaken. The numbers are each step’s place in the full path, and each one says what you can show when it is done.

  1. Step 1 of 16 Free

    The Autonomy Ladder: Which Level Is Your Workflow At?

    Place your current workflow on the ladder, so every later step is a move up one rung.

    Done whenYou can name your level and the one habit that keeps you on it.

  2. Step 3 of 16 Toolkit

    AGENTS.md and CLAUDE.md — concise repository context

    The instructions file is the cheapest lever on output quality; write it before you scale use.

    Done whenA reviewed CLAUDE.md or AGENTS.md under 150 lines with the real test commands.

  3. Step 5 of 16 Toolkit

    The artifact chain

    Intent, spec and plan are what you review instead of every line the agent writes.

    Done whenOne change shipped from an accepted intent.md, spec.md and plan.md.

  4. Step 7 of 16 Toolkit

    Test: give the session a feedback loop

    A single-command feedback loop lets the agent prove its own work before you look.

    Done whenThe agent runs the loop and reports exit status on every task.

  5. Step 8 of 16 Toolkit

    Protecting the oracle: checks the agent cannot edit

    Tests the agent can edit are not evidence; lock the checks it must not weaken.

    Done whenA hook or CI rule that fails when protected tests change.

Steps and “done when” lines: the developer reading path, 16 steps and 2 further reading, last updated 2026-10-12. The ladder pages are free; every other step is marked.

Read the whole path

Where your frontend work sits on the ladder.

Dan Shapiro’s levels describe how much of the work you hand to an agent. Each row shows what one frontend sweep looks like at a rung the scorecard can place you on.

Not sure which row is yours?

The AI Dev Score asks 25 questions in about 8 minutes, scores 6 sections and gives you a 30/60/90-day plan for the next rung. Free.

Take the AI Dev Score

No account, no card. An email only at the end, and only if you want the full playbook.

  1. 4

    The engineering team

    “leave for 12 hours, and check to see if the tests pass”

    One frontend sweepThe sweep is one task. You write the rule, the agent works until the type check and the tests pass, and you read the report. A change in an escalation class, such as authentication or money, still gets a named reader.

    Where the toolkit points
  2. 3

    The developer

    “Your life is diffs.”

    One frontend sweepThe agent renames the prop across the components and you read the whole diff, file by file.

    The valley, levels 2 and 3
  3. 2

    The junior developer

    “feels like you are done. But you are not done”

    One frontend sweepYou shape one component with the agent, a step at a time, and you check each render in the browser.

  4. 1

    The coding intern

    “you offload specific, discrete tasks to your AI intern”

    One frontend sweepYou hand over one component or one hook, and you paste the result into your own change.

Level names and quotations: Dan Shapiro, “The Five Levels: from Spicy Autocomplete to the Dark Factory”, 2026-01-23 (opens in a new tab). The ladder runs from Level 0 to Level 5 and the scorecard places you on 1 to 4. The “One frontend sweep” column is our illustration.

Your first week, planned.

  1. Day 1

    Generate your Setup Pack.

    Answer 9 questions about your stack and tools, including language, framework, package manager and test runner. Your agent runs INSTALL-PROMPT.md and checks each setting against what you have installed. About 20 minutes. The Setup Pack is included in the Toolkit.

  2. Days 2 to 5

    Hand over one real change.

    Pick a sweep you have been putting off, such as a prop renamed across the components. Write intent.md and spec.md, have the agent draft plan.md and accept it before any code changes, let the agent build and stop on the checks, then read the report.

  3. By day 30

    Decide with evidence.

    You have one change done the new way. If it did not change how you work, ask for a refund before day 30.

Buy the Toolkit

The Toolkit includes the guides, the Setup Pack and all eight books, for 12 months. $59.99 once for 12 months of access. Nothing renews on its own. 30-day refund on a first purchase.

Questions about this work.

Something else? Write to support@developertoolkit.ai.

Which tool has Tab completions?

Cursor documents Tab completion and Cmd+K inline edit in the editor itself. The comparison lists the Claude Code and Codex editor extensions with an agent panel and diffs, and names no Tab completion for either. Re-check a Cursor cell before you rely on it: the documentation was last read on 2026-08-28.

What about renaming a prop across forty components?

That is agent work, not autocomplete work. All three tools do autonomous multi-file editing with self-correction, so you hand over the sweep and read the diff and the results of the gates, not forty separate edits.

What do I read instead of the diff?

The evidence the change carries: the spec delta, the acceptance results, the strength of the oracle and runtime signals from a running build. A named person still reads the code for authentication, money, schema and migrations, changes to the tests themselves, and anything no check covers.

I only use one of the three tools.

That works. Where Claude Code, Codex and Cursor differ, the guides and recipes show a tab for each, so you read the one for your tool.

Keep the editor for shaping. Hand over the sweep.

Within the 30-day refund window, run one sweep this way on your own codebase.

$59.99 once for 12 months of access. Nothing renews on its own. 30-day refund on a first purchase.