whichguy/skill-craft

**Host-neutral portable skills monorepo.** Skills live under skills/ / and install into
Grok, Claude Code, Cursor, Codex, and Hermes skill directories via install.sh.
For setup and marketplace distribution across hosts, start with
docs/distribution.md.
Architecture (layers, binding, install honesty): docs/ARCHITECTURE.md.
This is **not** claude-craft.
| Repo | Role |
|------|------|
| **skill-craft** (this repo) | Skill source of truth and the marketplace for Claude, Codex, Grok and Cursor; install.sh for every host including OpenCode |
| **claude-craft** | Claude Code plugin marketplace (GAS, wiki, review suites, …) |
skill-craft vs claude-craft
- **skill-craft** is host-agnostic. There is no required root .claude-plugin. Skills follow an
agentskills.io-style unit: a directory with SKILL.md (and optional
prompts/, scripts/, references/).
- **claude-craft** is a Claude Code **plugin marketplace** (plugins with commands, agents, hooks).
Different packaging, different install path.
Use **skill-craft** for portable skill packages and their plugin distribution. Use
**claude-craft** for its separate product suites.
Skills in this monorepo
**23 skills.** Generated from skill frontmatter by scripts/sync-plugin-views.sh.
| Skill | Version | Purpose |
|-------|---------|---------|
| adversarial-review | 0.1.0 | Have an independent model try to break a proposed change and return its findings as testable hypotheses, each with the smallest experiment that would refute it. Use before… |
| architect | 0.1.2 | Design system architecture and make technology decisions. Uses a structured inline design or an available independent reviewer for comprehensive work. |
| ask-agent | 0.10.1 | A delegation skill, not an agent type. Ask native agents to work in the background, continue useful work in the main conversation, and incorporate their results when they return.… |
| backchain | 0.6.3 | Use when an implementation task needs a dependency-aware plan before coding: backward planner / backchain / precondition-first planning, dependency DAGs, elaborating incomplete… |
| c-plan | 0.1.2 | Resolve ambiguous user prompts by choosing whether to answer now, answer with assumptions, ask 1–2 high-value clarification questions, replan, or stop. Use when the best response… |
| compare-prompts | 0.1.3 | Compare two prompt versions (A vs B) by running both against a directory of test input files, then evaluating results on three dimensions in priority order: quality > tokens >… |
| improve | 0.4.2 | Use when a repository candidate needs a deliberate review-and-improvement loop: use recent Git history, make warranted changes, run meaningful checks, and repeat until two… |
| improve-agent | 0.1.3 | Use when Improve should run in a fresh native agent instead of this conversation: start one agent that runs /improve in its own context, keep useful parent work going, then… |
| improve-system-prompt | 0.1.2 | Benchmark and compare system prompt variants (V2/V2a/V2b/V2c) for Sheets Chat by running test scenarios through the real GAS-side ClaudeConversation pipeline. Tests both… |
| in-parallel | 0.1.1 | Use for "in parallel" or when one goal should be divided among independent native agents and returned as one verified aggregate. One coordinator owns decomposition, progress,… |
| plan-dispatcher | 0.6.4 | Use when executing or resuming an agreed dependency graph with one main dispatcher and parallel native workers: claim ready steps, preserve launch state, collect result evidence,… |
| plan-test | 0.1.3 | Generate comprehensive tests for code. Uses an inline strategy or an available independent test specialist for complex components. |
| prompt-align | 0.1.2 | Compare an agent or skill prompt against its test harness skill for phase-model, skip-condition, and wiring consistency. Reports mismatches and identifies which file is… |