Whetstone — AI skill for Claude Code
Small, sharp, agent-agnostic skills that keep an AI agent's work honest.
How to install Whetstone
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open jovesun-lab/whetstone and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Whetstone does
Small, sharp, agent-agnostic skills that keep an AI agent's work honest.
Alternatives in AI
- Maestro Companion — Quick execution for small tasks — minimal run lifecycle (start + done) with evidence recording 530 ★
- Ruby Nano Bots — Ruby Implementation of Nano Bots: small, AI-powered bots that can be easily shared as a single file, designed 126 ★
- Claude Prompt — Answer a few sharp, skippable questions about what you're building — and get back a clean, complete prompt rea 67 ★
README
Whetstone
*Small, sharp, agent-agnostic skills that keep an AI agent's work honest.*
A whetstone doesn't make a tool — it keeps its edge true. **Whetstone** is a growing collection of tiny open-source skills for AI agents: not frameworks, but **disciplines** that keep an agent's work on-track, high-quality, and easy to hand off.
What a Whetstone skill is
- Small and self-contained. One folder, one job. No build step, no install, no runtime.
- Agent-agnostic. Works as a plain prompt in any agent — Claude, OpenAI, Gemini, Cursor, a local model — and as a skill / slash command where those exist. Nothing assumes a particular platform; every capability degrades gracefully.
- Prompt-first, optional tooling. The discipline lives in plain language. Any helper script is optional, zero-dependency, and never required.
- Honest by construction. Checks are anchored to structure and substance, not vibes — a green light has to mean something.
- Plain words over jargon. If a reader has to learn a new token for an idea they already know, that's a defect, not concision.
The skills
| Skill | What it does | Works with any AI / model |
|---|---|---|
| Throughline | Task-Track & Handoff for AI agents on long, multi-session work — within a session, one ⭐️ MAIN goal anchor (frozen title, explicit verdict at close) in an emoji-tagged task list so reasoning doesn't drift; across sessions or agents, a plain-markdown handoff the next agent re-derives before trusting. | ✅ Yes — prompt-first, agent-agnostic |
| Session Measurement | Score an agent session by session and read the trend across versions — so a model / frame / skill change shows up as a real regression or improvement, not a vibe. Six honest counts per session, plain-markdown trend table (optional charts), any agent. | ✅ Yes — any agent |
| notify-me | Calls you back when an agent needs you — the moment it ge |
Related Skills
Model Identity Skill
Model-agnostic system prompt that aims to keep LLMs honest about their identity, capabilities, and knowledge c
Claude Toolkit
Small, honest Claude Code tools for the CLI and desktop app. /handoff mirrors your current session into a fres
Dev Coach
Stay sharp while your AI does the work — local MCP coach that teaches one concept per task.
Vibesubin
AI skills that keep your code honest, your secrets unleaked, and your repo from rotting. Built for people who
Sherwanis Claude Md
Drop-in AGENTS.md / CLAUDE.md that keeps AI coding agents honest: ask before assuming, keep it simple, change
Conus
A task manager your AI agents keep themselves. Claude Code, Codex and Gemini CLI create, start and close their
Related Agents
Backlog Manager
Use to keep the issue backlog HONEST and CONVERGING — the cross-issue counterpart to issue-maintainer's per-is
Small Business
COO and General Manager of the Small Business department, leading six employee skills that keep an owner-opera
Rowan
Lead. Plans Kernel sprint work, splits issues that are too big, and keeps Linear and docs/SCREENS.md honest. U
