Skill

craft-ace

Cost-optimized craft variant: Qwen3.6-35B-A3B handles implementation and reviews locally, Opus orchestrates, Sonnet as fallback. Requires LM Studio running. Targets ~45-60% API cost reduction vs base /craft.

npx claudepluginhub alexiolan/craft-skills --plugin craft-skills

Tool Access

This skill uses the workspace's default tool permissions.

Preview

Full design-first pipeline with Qwen3.6 as the primary implementer and reviewer. Opus orchestrates and handles integration. Sonnet serves as fallback for tasks the local model can't handle.

SKILL.md

Similar Skills

craft-local

Same as /craft but with local LLM review steps enabled. Use when you have LM Studio running and want LLM-assisted spec/plan/post-develop reviews. Preserves pre-1.1 craft behavior.

craft-skills

multi-ai-run

Routes agent roles to optimal AI models: Claude for planning/orchestration, Codex for code writing, Gemini for design/UI. Triggers on /multi-ai-run or model routing requests.

2 files

claude-impl-tools

craft-project-setup

Scaffolds CLAUDE.md and .claude/rules for Craft CMS projects (plugin, site, module, hybrid, monorepo) by detecting type from composer.json, .ddev/config.yaml, templates, and modules.

3 files

craftcms-claude-skills

Stats

Stars15

Forks0

Last CommitMay 5, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Craft ACE (Autonomous Claude Engine)

Full design-first pipeline with Qwen3.6 as the primary implementer and reviewer. Opus orchestrates and handles integration. Sonnet serves as fallback for tasks the local model can't handle.

Profile: claude+ace

Pipeline: Brainstorm -> Plan -> Develop -> Browser Test -> Report (same as /craft)

How this wrapper works

This wrapper writes a profile marker, then delegates to the canonical craft pipeline. The claude+ace profile triggers:

Qwen3.6 replaces Opus/Sonnet for spec and plan reviews (fixed two-pass with consensus consolidation)

Qwen3.6 handles all implementation tasks (data layer + UI)

Sonnet fallback for failed local-LLM UI tasks

Opus handles integration/wiring tasks only

Step 1: Write profile marker

echo -n "claude+ace" > .craft-profile

Step 2: Follow the craft pipeline

Read skills/craft/SKILL.md from this plugin and follow every phase exactly as written. The profile gating in that file ensures:

Step 1.10 (spec review): Qwen3.6 review loop replaces opus agent review

Step 2.4 (plan review): Qwen3.6 review loop replaces sonnet agent review

Step 3 (develop): Tasks routed to llm-implement.sh instead of Claude agents

The user input is: $ARGUMENTS

Pass the input through to the craft pipeline as if the user had invoked /craft directly.

Requirements

LM Studio running at ${LLM_URL:-http://127.0.0.1:1234}

Qwen3.6-35B-A3B (4-bit MLX) model loaded or available for auto-load

Does NOT require Codex CLI

Graph built on target project recommended (falls back to Glob)

Cost Model

Activity

Savings vs /craft

Reviews (spec, plan, post-develop)

100% (all local)

Data layer implementation (~40-50% LoC)

100% (all local)

UI implementation (~30-40% LoC)

~30-50% (Qwen3.6 first, Sonnet fallback)

Integration (~10-20% LoC)

0% (stays on Opus)

Total estimated savings

~45-60%

Craft ACE (Autonomous Claude Engine)

Full design-first pipeline with Qwen3.6 as the primary implementer and reviewer. Opus orchestrates and handles integration. Sonnet serves as fallback for tasks the local model can't handle.

Profile: claude+ace

Pipeline: Brainstorm -> Plan -> Develop -> Browser Test -> Report (same as /craft)

How this wrapper works

This wrapper writes a profile marker, then delegates to the canonical craft pipeline. The claude+ace profile triggers:

Qwen3.6 replaces Opus/Sonnet for spec and plan reviews (fixed two-pass with consensus consolidation)
Qwen3.6 handles all implementation tasks (data layer + UI)
Sonnet fallback for failed local-LLM UI tasks
Opus handles integration/wiring tasks only

Step 1: Write profile marker

echo -n "claude+ace" > .craft-profile

Step 2: Follow the craft pipeline

Read skills/craft/SKILL.md from this plugin and follow every phase exactly as written. The profile gating in that file ensures:

Step 1.10 (spec review): Qwen3.6 review loop replaces opus agent review
Step 2.4 (plan review): Qwen3.6 review loop replaces sonnet agent review
Step 3 (develop): Tasks routed to llm-implement.sh instead of Claude agents

The user input is: $ARGUMENTS

Pass the input through to the craft pipeline as if the user had invoked /craft directly.

Requirements

LM Studio running at ${LLM_URL:-http://127.0.0.1:1234}
Qwen3.6-35B-A3B (4-bit MLX) model loaded or available for auto-load
Does NOT require Codex CLI
Graph built on target project recommended (falls back to Glob)

Cost Model

Activity	Savings vs /craft
Reviews (spec, plan, post-develop)	100% (all local)
Data layer implementation (~40-50% LoC)	100% (all local)
UI implementation (~30-40% LoC)	~30-50% (Qwen3.6 first, Sonnet fallback)
Integration (~10-20% LoC)	0% (stays on Opus)
Total estimated savings	~45-60%