claudecraft

v0.2.2

Campaign-default improve (/improve-loop multi-cycle L1; bare /improve belongs to skill-craft): residual×2 complete, git learnings digest, multi-model replan, ambient launch-dirt tolerance. Normative law at law/improve-loop/; thin /claudecraft:improve continuous host optional. Ship set is B monolith + scripts.

Claude Code2 Skills

By whichguyLicense: MIT2 GitHub starsUpdated 3 weeks ago

Directory evidence

Runtimes: Claude Code
Parsed components: 2 skill or MCP entries
Source updated: Sep 26, 2026
Manifest status: Canonical path parsed

The directory validates manifest shape and source location. It does not execute the plugin or provide a security endorsement. Review the indexing methodology →

Install claudecraft for Claude Code

Installs for the current user

claude plugin marketplace add IchenDEV/agent-plugin-mkt
claude plugin marketplace update agent-plugin-marketplace
claude plugin install claudecraft@agent-plugin-marketplace

Paste and run these commands in a terminal with Claude Code. They add and refresh the PluginsMP catalog, then install this plugin.

The installer fetches third-party code from the source repository shown on this page. This directory validates manifest structure and source location, but does not perform a security audit; review the manifest, components, and source before installing.

Get the source manually

$ git clone https://github.com/whichguy/claude-craft

Clone the source repository, then follow its setup instructions to add the plugin to a compatible client. The plugin root is plugins/claudecraft/.

Plugin files

plugins/claudecraft/

├── .claude-plugin/plugin.json
├── skills/improve/SKILL.md
└── skills/improve-loop/SKILL.md

Included Skills2

improveskills/improve/SKILL.md

Continuous evidence-led improvement for any git repo (host-agnostic driver): NL parse, worktree lifecycle, native or host-goal outer loop, caps that always reintegrate. Each inner step is one improve-loop cycle. Plugin invoke "/claudecraft:improve …"; one-shot only → improve-loop skill instead.

improve-loopskills/improve-loop/SKILL.md

Run a worktree-isolated improvement campaign until terminal, then emit one end report: default is autonomous multi-cycle in a single invoke (execute, test, learn, multi-model replan, commit, loop). Use --once for a single gated cycle; use --product / product / ux / integration / roadmap for product/UX seed mode (limitations + deferred P2 + operator surfaces, not residual-empty-only). Invoke for "/improve-loop <target>", "/improve-loop --once <target>", "/improve-loop --product <target>", "/claudecraft:improve-loop <target>", "improve this project iteratively", or "run a tested improvement cycle". Not for bare "/improve" or "improve these changes": those belong to the separate skill-craft improve skill.

Plugin manifests1

plugins/claudecraft/.claude-plugin/plugin.json

{
  "name": "claudecraft",
  "version": "0.2.2",
  "description": "Campaign-default improve (/improve-loop multi-cycle L1; bare /improve belongs to skill-craft): residual×2 complete, git learnings digest, multi-model replan, ambient launch-dirt tolerance. Normative law at law/improve-loop/; thin /claudecraft:improve continuous host optional. Ship set is B monolith + scripts.",
  "author": {
    "name": "whichguy",
    "url": "https://github.com/whichguy"
  },
  "homepage": "https://github.com/whichguy/claude-craft",
  "repository": "https://github.com/whichguy/claude-craft",
  "license": "MIT",
  "keywords": [
    "claude-code",
    "improve-loop",
    "improve",
    "iteration",
    "tested-improvement",
    "worktree"
  ],
  "dependencies": []
}