judgment-over-velocity
orchimada/judgment-over-velocity
A Claude skill for deciding what NOT to build — faster. Four kill-gates: spreadsheet test, 80/20 cut, figure/ground, evidence probes.
Instalação
npx convez-ai orchimada/judgment-over-velocity1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
Judgment over Velocity
The job isn't to build faster. It's to decide what not to build — faster.
A Claude skill that turns Claude into a structured product-judgment engine. Give it a product idea, a backlog, a "should we buy a tool?" question, or any CPO-level problem — and instead of cheerfully helping you build it, the skill classifies the problem first and runs it through four kill-gates.
It is explicitly willing to answer "don't build anything" — the most common right answer and the rarest one product teams hear.
Made by Artem Domozhakov · How it works (long version)
The four gates
Problem/Idea → [1. Spreadsheet test] → [2. 80/20 structural cut] → [3. Figure/ground] → [4. Evidence probes] → Verdict + cheapest test
↓ kill ↓ kill ↓ kill ↓ kill
epitaph epitaph epitaph epitaph
- The spreadsheet test — most product problems are discipline problems wearing tooling costumes. Could a competent operator solve this with a table, an operating rhythm, and clear decision rights? Field calibration: 4 of 6 core CPO problem clusters fail this gate.
- The 80/20 structural cut — every option framed in commercial terms; the field scored as a whole. The no-list is the deliverable: each cut gets a one-line reason a stakeholder could repeat.
- Figure/ground analysis — products judged by the behavior they install, not the features they list. What becomes scarce when your capability makes something abundant? Which assumptions are load-bearing — and melting?
- Evidence probes & adversarial loop — probes, not opinions. Demand signals ranked: expensive workarounds > apologies > nostalgia > stated feature requests. Survivors get the cheapest disconfirming test, with the kill-condition stated in advance.
Each gate can kill the idea — killing early is the point. Killed ideas get epitaphs (what killed them, what would resurrect them).
Install
Claude Code
# personal — available in every project
git clone https://github.com/orchimada/judgment-over-velocity.git /tmp/jov \
&& cp -r /tmp/jov/judgment-over-velocity ~/.claude/skills/
# or project-level — shared with your team via git
cp -r /tmp/jov/judgment-over-velocity .claude/skills/
Restart Claude Code (or start a new session) and verify: ask "do you have the judgment-over-velocity skill?"
claude.ai (web & desktop)
- Download
judgment-over-velocity.skill(it's a zip of the skill folder). - Settings → Capabilities → Skills → upload the file.
The skill activates automatically whenever you ask Claude to evaluate an idea, prioritize a backlog, or judge a build-vs-buy question.
Try it on
- "We're thinking of building an internal dashboard to track product KPIs across teams. Should we?"
- "Here are the 14 items on our Q3 roadmap — help me prioritize."
- "Sales wants us to buy a customer-feedback platform. We get ~30 feedback items a quarter."
- "Evaluate this idea: AI agent that writes PRDs from Slack threads."
Every run ends in a verdict
## Verdict
**Classification:** DISCIPLINE | TOOLING-EARNED | HYBRID
**The 20%:** the 1–3 moves that carry the impact, each with its business number
**The no-list:** everything cut, one defensible line each
**Behavior installed / scarcity owned:** one line
**Riskiest assumption:** the load-bearing assumption most likely to be false
**Cheapest disconfirming test:** what to run, what it costs, what result kills it
**Epitaphs:** killed ideas — what killed each, what would resurrect it
What's inside
| File | Contents |
|---|---|
judgment-over-velocity/SKILL.md | The four-gate pipeline, verdict format, and tone rules ("never soften a kill into a maybe-later") |
references/spreadsheet-test.md | Full diagnostic: three classification questions, the six-cluster CPO calibration map, HYBRID handling, named anti-patterns (tool-as-avoidance, dashboard theater, pilot purgatory) |
references/figure-ground.md | Five instruments: behavior-installed test, abundance→scarcity table, invisible-assumptions audit, seams between domains, the rear-view-mirror check |
judgment-over-velocity.skill | The same skill folder zipped for claude.ai upload |
License
MIT © 2026 Artem Domozhakov