VividKit Crystal clear AI coding
VI
Choose Your Setup Get Started with AgentKit Continue with ClaudeKit Migrate from ClaudeKit to AgentKit Clean cutover from ClaudeKit What Is AgentKit? How AgentKit Works AgentKit Kits, Skills, Agents & Hooks AgentKit Runtime Support AgentKit Desktop App AgentKit Configuration AgentKit CLI Quick Reference AgentKit Skills Cheatsheet AgentKit Workflows AgentKit Statusline Update AgentKit CLI & Kits AgentKit Helper Clean reinstall of AgentKit AgentKit Troubleshooting What is ClaudeKit? ClaudeKit CLI Guide Command Finder Commands Cheat Sheet Workflow Recipes CLI Quick Reference Claude Mechanics Permission Modes ClaudeKit Hooks Coexistence ClaudeKit x Codex CK → Other coding tools How ClaudeKit Works Inside ClaudeKit Finding Your Unknowns Session Recovery Fix from Logs UI/UX Guide Accounts Switcher IDE Config Remote Control Codex App System Diagramming UI Review Gate Promotions Support VividKit

Quick Ref / Command

Engineer Kit

/ak:bootstrap

New project from idea to running code

Bootstraps web, mobile, CLI, API, MCP, WebMCP, and library projects from a locked scope straight to implementation via ak:cook, then code review, UX/AX enhancement for web surfaces, and optional release with live verification. There is no separate planning phase.

01

Lock the contract

02

Set up the repository

03

Research, stack, design

04

Cook directly

Rule 01

Bootstrap orchestrates end to end; it never implements code directly or adds a separate ak:plan phase.

Rule 02

The opening contract is mandatory in every mode; modes change execution and approval behavior only after it is concrete.

Rule 03

Full requested scope is the default; --yagni is the only opt-in to challenge and cut unnecessary scope.

Rule 04

A project is done only when its definition of done is met, including live verification when it names a live target.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Lock the contract Before Git, research, design, or scaffolding, capture outcome, constraints, non-goals, acceptance criteria, and definition of done, checked against the project-brief checklist for each surface. --auto decides gaps as recorded assumptions; --ask interviews through ak:advise until you confirm.
  2. 2 Set up the repository Inspect the destination and Git state, reuse an existing project and its branches, and initialize Git only for a new project when the setup includes it. Create a remote repository through ak:github only when the brief asks.
02

Work

Agent / skill execution

  1. 3 Research, stack, design Run the mode reference for research and stack choice, routing each surface to its skill. Visual projects use ak:frontend-design when art direction matters; CLI, API, MCP, and library projects design their interface contract instead of wireframes.
  2. 4 Cook directly Pass the contract, assumptions, and research, stack, and design outputs to /ak:cook with the mode flag. Cook sizes its own plan and runs the tests; bootstrap does not call ak:plan.
03

Verify

Gate, review, validation

  1. 5 Code review Run ak:code-review on the new project (codebase, or --pending, PR, or commits for work in an existing project). Blocking findings go back through cook, then review again with the same flags.
  2. 6 UX and AX for web When the project has a website or web app, run ak:enhance-ux-ax --auto after review passes. Projects without a web surface skip this phase.
04

Close

Report, handoff, artifact

  1. 7 Release and verify live When the definition of done names a live target, release there (staging first when defined) and verify the way a real user or client reaches it: browser for web, clean install for a CLI, a real MCP client for a server.
  2. 8 Docs, onboard, report Write docs and onboarding, report the outcome with unresolved questions last, and run /ak:journal unless --skip-journal or preferences disable it.

Syntax, arguments, subcommands

How to invoke the skill: syntax, positional arguments, shared options, then each subcommand with its own syntax and outcome.

Syntax

/ak:bootstrap [requirements] [--auto|--full|--fast|--parallel] [--ask] [--ultra] [--yagni] [--skip-journal]

Arguments

[requirements] Required

Project requirements

Natural-language product outcome, surfaces, constraints, non-goals, acceptance criteria, and definition of done, including any live target such as production or a registry.

/ak:bootstrap "Create a private incident-response dashboard with SSO, an audit log, and no billing in v1"

Shared options

--auto

Automatic (default)

Default mode. Decides open gaps as recorded assumptions and continues without routine approval pauses, stopping only at authorization boundaries. Does not grant commits, publication, or deployment the brief does not ask for.

--full

Full interactive

Refines requirements with you, then research, stack choice, and design with pauses on material decisions before cook.

--fast

Fast

Uses a lean research and design path, then /ak:cook --fast. Does not reduce code review or verification.

--parallel

Parallel

Runs research and implementation with multiple agents through /ak:cook --parallel. Does not make overlapping edits safe.

--ask

Interview to lock scope

Not a mode. Runs the opening gate as an ak:advise interview until outcome, scope, constraints, non-goals, acceptance criteria, and definition of done are confirmed, then continues in the selected mode.

--ultra

Ultra code review

Runs the code review phase with the best-of-5 verifier mode (/ak:code-review codebase --ultra). Combines with every mode; bootstrap itself does not fan out.

--yagni

Cut unneeded scope

Challenge and cut requested work that is not needed for the stated outcome. Without this flag, bootstrap builds the full requested scope.

--skip-journal

Skip journal

Skip only the automatic /ak:journal step at completion. Explicit journal commands still work.

Composable flags

How options work together on one invocation. This does not replace the option cards above.

--ask runs the opening gate as an ak:advise interview, then continues in the selected mode.

--ultra makes the code review phase run /ak:code-review codebase --ultra and combines with every mode, including --parallel.

--yagni passes through to cook and code review.

--skip-journal only affects the final journal step.

Modes / Routes

--auto

Automatic (default)

Research:
Automatic research and stack choice
Review:
Authorization boundaries only
Validate:
Cook --auto; code review
--full

Full interactive

Research:
Refined requirements, then research
Review:
Missing material decisions
Validate:
Cook interactive; code review
--fast

Quick bootstrap

Research:
Lean research path
Review:
Missing material decisions
Validate:
Cook --fast; code review
--parallel

Multi-agent

Research:
Parallel research
Review:
Missing material decisions
Validate:
Cook --parallel; code review

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Default autonomous run Recommended 
/ak:bootstrap "MCP server and CLI for our billing API"
Use when:
Use when the brief is clear enough for the agent to decide remaining gaps itself.
Expected:
Locks the contract with recorded assumptions, designs the CLI commands and MCP tool schemas, cooks with --auto, runs code review, and stops only at authorization boundaries.
Interview first 
/ak:bootstrap "Redesign our marketing site" --ask
Use when:
Use when the scope is vague and you want to lock it in an interview before any build.
Expected:
Interviews through ak:advise until outcome, scope, and definition of done are confirmed, builds the new site in its requested location without overwriting the old one, then runs review and ak:enhance-ux-ax --auto.
Fast SaaS scaffold 
/ak:bootstrap "Build a SaaS dashboard with auth" --fast
Use when:
Use when a new project needs quick setup but still must pass the opening contract and review.
Expected:
Captures the contract, asks only about decisions that would change the product, takes the lean research path, cooks with --fast, and runs code review.
Parallel build 
/ak:bootstrap "E-commerce platform with Stripe" --parallel
Use when:
Use when the project is large enough for multi-agent implementation.
Expected:
Keeps the opening contract, runs parallel research, passes the contract to /ak:cook --parallel, and reviews the combined result before UX/AX enhancement.
Full interactive 
/ak:bootstrap "CLI tool for invoices" --full --skip-journal
Use when:
Use when you want to approve requirements, stack, and design yourself and skip the automatic journal.
Expected:
Refines requirements with you, pauses on material decisions, cooks interactively, runs code review, and reports that the journal was skipped by flag.
Ultra review 
/ak:bootstrap "Internal analytics portal" --auto --ultra --yagni
Use when:
Use when the code review should use the best-of-5 verifier and scope should be challenged.
Expected:
Cuts scope not needed for the outcome, cooks in --auto mode, and runs /ak:code-review codebase --ultra, including the re-review after UX/AX enhancement.

Handled Scope

  • Lock the contract
  • Set up the repository
  • Research, stack, design
  • Cook directly
  • Code review
  • UX and AX for web

Next