Claude Toolkit — AI skill for Claude Code
Small, honest Claude Code tools for the CLI and desktop app.
How to install Claude Toolkit
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open waltwildwest/claude-toolkit and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Toolkit does
Small, honest Claude Code tools for the CLI and desktop app. /handoff mirrors your current session into a fresh one (same model, effort and permissions). /route sends delegated work down to the cheapest capable model, keeps judgment and review on yours, and reports the savings against honest baselines.
Alternatives in AI
- LongHorizon Harness — The long-horizon computer-use harness 1.3k ★
- Codex Skill — by klaudworks - Enables users to prompt codex from claude code 914 ★
- Maestro Companion — Quick execution for small tasks — minimal run lifecycle (start + done) with evidence recording 530 ★
README
claude-toolkit
Small, honest tools for [Claude Code](https://claude.com/claude-code) that work in both the CLI and the desktop app.
[](./LICENSE) [](https://claude.com/claude-code)
Tools I actually use every day. **Each one lives in its own repo, stands on its own merits, and installs on its own** — this repo is the marketplace index that ties them together. No framework, no lock-in, nothing phones home.
That one-plugin-per-repo shape is deliberate. The unit of install, enable and disable in Claude Code is the *plugin* — mega-bundles force you to adopt (and context-load) a hundred skills to use one. Here you add the marketplace once and install exactly the tools you want; each repo's README goes deep on the process, the capabilities, and the why of that single tool.
The tools
| Tool | One line | Why it exists |
|---|---|---|
| **handoff** (+ pickup) | Pass a task to a fresh session that mirrors your model, effort and permission mode | Long sessions rot; restarting shouldn't cost you the decisions, dead ends and verify steps the old session already paid for |
| **route** | Cost-aware delegation: size → cache → cheap-under-expensive-review → honest savings report | Most agent work is grunt work; route it down the ladder and prove the savings against real baselines instead of vibes |
| **re-searcher** | A persistent research vault with recall-first answers and quote-verified claims | Research that evaporates into a chat summary gets re-derived forever; claims should come back dated, sourced and falsifiable |
Install
Add the marketplace once, then install only what you want:
/plugin marketplace add waltwildwest/claude-toolkit
/plugin install handoff@claude-toolkit
/plugin i
Related Skills
Whetstone
Small, sharp, agent-agnostic skills that keep an AI agent's work honest.
Ashlr Genome Init
Initialize a .ashlrcode/genome/ in the current project so ashlr__grep can route through genome RAG for ~-84% t
Frontier Handoff
Package the current stuck situation (code, problem, what was tried, environment, ask) into ONE self-contained
Model Route
Recommend and apply the right model tier for the current task
Modal Claude Agent SDK Python
This package wraps the Claude Agent SDK to execute AI agents in secure, scalable Modal containers. It provides
Agent Leash
LEASH-8: an 8-domain control model for AI agents with delegated authority. Scorecard, approval-design checklis
Related Agents
Mid Level Engineer
Implements a single, narrowly scoped coding task delegated by the executor. Takes a specific file, function, o
Courier
Close out a delivered effort. Packages the output, generates recipient-specific briefs, and writes the permane
Dpo Review
Fresh-eyes data-protection & consent review for changes touching marketing sends, consent/suppression, or cust