judgment-distiller — Claude Code skill: decisions and setup
judgment-distiller is a Claude Code skill that turns an accept-or-reject decision and its rationale into a project-scoped learning candidate. It keeps observations separate from broader principles.
MIT ★ 529 updated 2026-11-17
npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a claude-codeWhat judgment-distiller does
judgment-distiller records why a person accepted or rejected a change and shapes that reasoning into a candidate for project learning. It needs the decision and rationale, the finding’s context, and evidence from before and after the change.
Candidates default to project scope and need review before promotion. One judgment cannot establish a universal design rule. If evidence or professional review is missing, the skill returns needs-evidence; a withdrawn judgment remains historical evidence and produces no learning candidate.
What it helps with
- record why a change was accepted
- capture the rationale for rejecting a change
- separate a project observation from a general principle
- flag missing evidence before proposing a learning candidate
Example prompts for the agent
- Record why we accepted this change and turn the rationale into a project learning candidate.
- Review this rejected change, preserving its reason and finding context.
- Compare the evidence from before and after the change, then identify what is missing.
How to install judgment-distiller
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a claude-code2Manually: put the files in place
~/.claude/skills/judgment-distiller/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/judgment-distiller/
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a codex2Manually: put the files in place
.agents/skills/judgment-distiller/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a cursor2Manually: put the files in place
.cursor/skills/judgment-distiller/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install ShuaiMXu/happyhands skills/product-design/judgment-distiller2Manually: put the files in place
.github/skills/judgment-distiller/SKILL.md1Skills CLI (Vercel)
npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a gemini-cli2Platform CLI
gemini skills install https://github.com/ShuaiMXu/happyhands --path skills/product-design/judgment-distillerManually: put the files in place
.devin/skills/judgment-distiller/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about judgment-distiller
How do I install judgment-distiller in Claude Code?
Run npx skills add ShuaiMXu/happyhands --skill judgment-distiller -a claude-code. The skill file is at ~/.claude/skills/judgment-distiller/SKILL.md.
What does judgment-distiller do?
It turns the rationale and context behind an accepted or rejected change into a candidate for project learning. It distinguishes observed evidence from inferred principles and requires review before promotion.
Can judgment-distiller derive a universal rule from one decision?
No. A single judgment cannot establish a universal rule, so candidates default to project scope. When evidence or professional review is missing, it returns needs-evidence.
Source text
SKILL.md · MIT
--- name: judgment-distiller description: "Turn an accept or reject decision and its rationale into a scoped project-learning candidate." --- # Judgment Distiller Preserve why a human accepted or rejected a change without over-generalizing. ## Required evidence - decision and reason - finding context - before and after evidence ## Decisions - Default the candidate to project scope - Separate observation from inferred principle - Require review before promotion - A withdrawn judgment is historical evidence only: return `needs-evidence` with no learning candidate. Retain its original decision and reason for audit; start a new review to reassess it. ## Boundary One judgment cannot establish a universal design rule. Return structured evidence, findings, proposed updates and limitations. Use `needs-evidence` when required inputs or professional review are missing. Never invent an Experience Score or promote a project observation into a universal rule.
© ShuaiMXu · MIT · ShuaiMXu/happyhands