Official stable reference · docs.claudekit.cc

ClaudeKit Engineer

The complete agent & skill reference, tracking the official stable release line. Beta channel excluded by design.

v2.20.0 stable Released 2026-06-18 11 stable releases tracked
14agents
88skills
5new in v2.20.0
11stable releases

Release channels

This document follows the Stable channel. Each skill carries a chip showing the release it first shipped in — the latest, NEW · v2.20.0, is highlighted.

Stablelatest

main branch — Production-ready, fully tested — ck update

Betabeta

dev branch — Pre-release, may have rough edges — ck update --beta (excluded here)


Official changelog

Stable releases, newest first. Pre-release work on the beta (dev) channel is omitted.

v2.20.0

2026-06-18
  • New skills: ghpm (GitHub project management — issues, Projects boards, handoffs), html-video (HTML/CSS/JS → MP4), vibe (full vibe-coding pipeline), review-pr (formal gh review with --reply + anti-ai-slop reference).
  • chrome-profile — real Chrome profile workflow, replacing the deleted chrome-devtools.
  • ck:plan — drops time estimates; adds AgentWiki publish flag + HTML/GitHub output modes.
  • git PR merge automation; ship publishes journals to AgentWiki; watzup cross-branch handoff scanner with ranked next steps.
  • shopify modernized to current shopify.dev (GraphQL-first + safe init wrapper); brainstorm gains --html wiki & problem-first modes.
  • Hooks — workflow artifact gate, canonical wiring overhaul, managed-hooks manifest for deterministic self-heal. Human MCP disabled by default; code-review red-team wording made policy-safe.

v2.19.0

2026-05-13
  • Skill gates — scout-first, exact-requirements, and no-side-effects enforced on shipped skills.
  • Generated context hooks disabled by default.

v2.18.0

2026-05-12
  • New skill: tech-graph (vendored fireworks-tech-graph) — semantic code analysis.
  • ck:predict --chain reason/probe; ck:scenario iterative saturation; ck:security red-team personas; ck:autoresearch reframed as a family router.
  • agent-browser refactored to a discovery stub; mermaidjs-v11 SVG layout reference routed to tech-graph.
  • Removals: chrome-devtools and template-skill deleted; kanban + mcp-management duplicates absorbed into use-mcp.

v2.17.0

2026-04-28
  • New skills: agentize (codebase → CLI + MCP server), ai-multimodal (multi-provider image routing), cti-expert, excalidraw, graphify, show-off, xia; stitch per-project isolation.
  • worktree --base to override the auto-detected base branch; deep + TDD plan-cook workflows.
  • Skill schema contract + format-compliance gate; 82 skill descriptions hardened.

v2.16.0

2026-04-03
  • Statusline overhaul — config-driven modular architecture, quota reset countdown, custom color themes, user-defined layout, cache-first startup.
  • Session-state moved to global storage (no project-dir pollution); per-project .ckignore overrides for scout-block.
  • ai-multimodal env key resolution + 5xx retry; stitch SDK fixes + auto quota tracking; worktree --no-prefix / --help; useGemini default → false.

v2.15.0

2026-03-29
  • worktree --no-prefix flag to skip branch prefix and preserve case.
  • (v2.15.1: release tooling only — beta workflow + Discord notification fixes.)

v2.14.0

2026-03-28
  • New skills: retro, llms, deploy, project-organization, loop, scenario, predict, security, stitch, security-scan, ship, design (consolidated logo/CIP/slides/banner/icon).
  • fix v2.0 — structured Scout → Diagnose → Assess → Implement → Verify pipeline; team v3.0.0 on the current Claude Code Agent API.
  • code-review always-on adversarial review (4 input modes, scope gating); preview --html with diff / plan-review / recap modes + theme toggle.
  • Hooks — session-state persistence + strategic compaction + structured diagnostics logging; 3-tier eval infrastructure for skills and agents.

v2.13.0

2026-02-25
  • ck: namespace prefix — every skill name is prefixed with ck: to avoid collisions with Claude Code built-ins.
  • argument-hint field added to all skills for better discoverability.
  • Namespace cleanup and cross-reference fixes across the kit.
  • chrome-devtools evaluate fix.

v2.12.0

2026-02-05
  • Agent Teams v2.1.0 — rewritten as a CK-native imperative execution engine. Auto-executing templates, event-driven hooks (TaskCompleted, TeammateIdle), agent memory persistence.
  • Portable manifest — Codex AGENTS.md.codex/agents/ migration support.
  • /ck:plan red-team — adversarial plan review with hostile reviewer lenses.
  • Commands → Skills — every command is now a skill; /ck: syntax unchanged.
  • New skills: team, plan, bootstrap, test, project-management, mintlify. New agent: Code Simplifier.
  • Native Task orchestration for fix / code-review / scout / debug; skill-creator Skillmark support; Chrome DevTools headless detection + OAuth.

v2.11.0 – v2.11.3

2026-02-20 → 21
  • Always-on diagnostics logging — crash wrappers on every hook for automatic error capture.
  • Scout-block hardening — broad glob detection (**/*.ext), heredoc-aware path extraction, compound-command & allowlist edge cases, path normalization.
  • Windows: reduced terminal freeze from excessive process spawning.
  • Added hotfix: commit type; Discord embed field-limit & empty-changelog guards.

v2.10.0

2026-02-05
  • Focused Reader 2.0 — redesigned markdown-novel-viewer.
  • Plan validation propagation — auto-propagate validation decisions to phase files.
  • Visual generation modes/ck:preview --explain --slides --diagram --ascii.
  • cook-after-plan-reminder hook; code-review edge-case scouting; language-aware naming hook; monorepo worktree paths.
Releases prior to v2.10.0 live on the GitHub Releases page. Beta (dev-channel) builds are intentionally excluded from this stable reference.

Agents

14 specialized agents

Orchestrated automatically by skills. code-simplifier v2.12.0 was the most recent agent added.

Development & Implementation
planner

Research, analyze, create implementation plans before coding

fullstack-developer

Execute implementation phases with strict file ownership

debugger

Root cause analysis, log investigation, issue diagnosis

tester

Test execution, coverage analysis, quality validation

Quality & Review
code-reviewer

Security audits, performance analysis, code quality

code-simplifierv2.12.0

Autonomous code refinement for clarity and maintainability

Documentation & Management
docs-manager

Technical documentation, API docs, architecture guides

project-manager

Progress tracking, cross-agent coordination, status reports

journal-writer

Document failures and setbacks with brutal honesty

git-manager

Conventional commits, security scanning, token-optimized

Creative & Research
ui-ux-designer

Award-winning UI with Three.js, responsive layouts

brainstormer

Explore approaches, challenge assumptions, debate decisions

researcher

Multi-source research, documentation analysis


Skill catalog

Invoke with /ck:skill-name. Chips show the release a skill first shipped in.

brainstorm

Trade-off analysis with brutal honesty; --html wiki + problem-first modes (v2.20.0)

vibeNEW · v2.20.0

Full vibe-coding pipeline end-to-end

planv2.12.0

Transform requirements into executable plans; HTML/GitHub output, AgentWiki publish (v2.20.0)

predictv2.14.0

Pre-implementation debate; --chain reason/probe (v2.18.0)

scenariov2.14.0

Generate edge cases & test scenarios; iterative saturation (v2.18.0)

cook

Feature implementation workflow; deep + TDD modes (v2.17.0)

testv2.12.0

Run test suites, coverage analysis, build verification

code-review

Verification gates, red-team review, 4 input modes (v2.14.0)

review-prNEW · v2.20.0

Formal gh PR review (--reply) + anti-ai-slop; duplicate/product gates

shipv2.14.0

Unified ship pipeline (branch → PR); AgentWiki journals (v2.20.0)

fix

Bug fixing with intelligent routing (v2.0 Scout→Diagnose→Assess→Implement→Verify)

debug

Root cause investigation framework

loopv2.14.0

Autonomous optimization loop against a metric

securityv2.14.0

STRIDE + OWASP security audit; red-team personas (v2.18.0)

security-scanv2.14.0

Security vulnerability / secret scanner

ask

Answer technical and architectural questions

research

Multi-source validation before implementation

autoresearch

Autoresearch family router (v2.18.0)

problem-solving

Systematic approaches when stuck

sequential-thinking

Numbered thought sequences for complex problems

coding-level

Set explanation depth (beginner → expert)

scout

Fast parallel codebase scouting

repomix

Pack repos into AI-friendly context files

gkg

GitLab Knowledge Graph semantic analysis

graphifyv2.17.0

Build queryable knowledge graphs from code/docs

tech-graphv2.18.0

Semantic code analysis — go-to-definition, find-usages, impact analysis

teamv2.12.0

Agent Teams — parallel multi-session collaboration (v3.0.0)

plans-kanban

Integrated plans dashboard in the CLI UI

project-managementv2.12.0

Task tracking, plan status, session bridging

ghpmNEW · v2.20.0

GitHub project management — issues, Projects boards, handoffs, gh automation

agentizev2.17.0

Convert codebases into CLI + MCP server

frontend-design

Build memorable web interfaces; upstream design guidance (v2.20.0)

frontend-development

React patterns, Suspense, state management

ui-styling

Tailwind patterns, responsive layouts, dark mode

ui-ux-pro-max

Production-ready UI with research-backed patterns

web-design-guidelines

Web Interface Guidelines compliance

react-best-practices

React/Next.js performance optimization from Vercel

stitchv2.14.0

Google Stitch AI design generation; --project-name isolation (v2.17.0)

threejs

3D web experiences with WebGL/WebGPU

shader

GLSL fragment shaders for procedural graphics

remotion

Programmatic video creation in React

designv2.14.0

Consolidated logo / CIP / banners / icons / slides

backend-development

Node.js, NestJS, security, testing patterns

better-auth

TypeScript auth — OAuth, 2FA, passkeys, multi-tenant

databases

Schema design, query optimization, migrations

payment-integration

Stripe, PayPal, LemonSqueezy, SePay

web-frameworks

Next.js + Turborepo monorepos

tanstack

TanStack Start / Form / AI for full-stack React

mobile-development

React Native, Flutter, Swift/SwiftUI, Kotlin

shopify

Shopify apps, GraphQL-first API, checkout extensions (modernized v2.20.0)

google-adk-python

Google AI Development Kit for Python agents

deployv2.14.0

Multi-platform deployment with auto-detection

devops

CI/CD, deployment, infrastructure automation

worktree

Isolated git worktrees; --base (v2.17.0), --no-prefix/--help (v2.16.0)

web-testing

Playwright, Vitest, k6 testing

chrome-profileNEW · v2.20.0

Real Chrome profile browser automation (replaces chrome-devtools)

agent-browser

AI-optimized browser automation (discovery stub, v2.18.0)

docs

Manage project documentation (init, update, summarize)

docs-seeker

Find and retrieve external documentation

llmsv2.14.0

llms.txt generator (llmstxt.org spec)

mintlifyv2.12.0

Build and deploy Mintlify docs sites

document-skills

PDF, DOCX, PPTX, XLSX processing

mcp-builder

Create MCP servers (Python FastMCP / TypeScript)

use-mcp

Discover, select, execute MCP capabilities (absorbed mcp-management, v2.18.0)

ai-multimodalv2.17.0

Multi-provider image routing; Gemini vision, audio, docs

ai-artist

Generate images via Nano Banana with curated prompts

media-processing

FFmpeg, ImageMagick, background removal

html-videoNEW · v2.20.0

HTML/CSS/JS templates → local MP4 via nexu-io/html-video

preview

View files or generate visual explanations & diagrams

show-offv2.17.0

Stunning self-contained HTML showcase pages

excalidrawv2.17.0

Excalidraw diagrams — architecture, data flow

mermaidjs-v11

Diagram creation with Mermaid.js v11

context-engineering

Monitor and optimize token consumption

find-skills

Discover and install agent skills

skill-creator

Create custom skills with eval-driven iteration

copywriting

Conversion copy formulas and templates

markdown-novel-viewer

Book-like markdown reading experience

cti-expertv2.17.0

Cyber threat intelligence / OSINT analysis

git

Git operations with conventional commits; PR merge automation (v2.20.0)

watzup

Cross-branch handoff + session wrap-up (v2.20.0)

journal

Write development journal entries

retrov2.14.0

Data-driven sprint retrospectives

bootstrapv2.12.0

Initialize new projects with spec-driven development

project-organizationv2.14.0

File / directory structure standardizer

xiav2.17.0

Port / extract / compare features from another repo


Recommended workflows

Common command paths for daily engineering work.

Standard dev flow

/ck:brainstorm → /ck:predict → /ck:plan → /ck:scenario → /ck:cook → /ck:test → /ck:code-review → /ck:ship → /ck:watzup

Large features (parallel)

/ck:brainstorm → /ck:plan → /ck:team cook → /ck:team code-review → /ck:ship

Bugfix pipeline

/ck:scout → /ck:debug → /ck:fix → /ck:test → /ck:code-review

Optimization

/ck:scout → /ck:loop → /ck:code-review → /ck:ship