SkillProof tested Claude skills

Oracle Challenge

Structured critical-thinking framework to pressure-test a decision before you commit.

Autor: martinffx · martinffx/atelier

Działa po konfiguracji ★ 6.4/10

Oracle Challenge — Structured critical-thinking framework to pressure-test a decision before you commit.

Co robi ten skill

Runs a 4-step challenge framework (assumptions, evidence, alternatives, risk) against a stated approach or decision. Triggers on requests to question assumptions, play devil's advocate, or validate a technical/architectural decision before committing to it.

Raport z testu

The framework text is a solid critique checklist, but Step 1 and Step 3 literally invoke '@agent-oracle' and an MCP sequential-thinking tool that a verbatim skill-folder copy does not include — confirmed by installing only skills/oracle-challenge and finding no ~/.claude/agents/oracle.md present, so Claude has to run the framework itself instead of delegating as documented.

Testowano: 2026-07-15 · Claude Code 2.x (agent harness)

Instalacja

git clone https://github.com/martinffx/atelier
mkdir -p ~/.claude/skills
cp -r atelier/skills/oracle-challenge ~/.claude/skills/oracle-challenge

Komendy i przykładowe prompty

  • /oracle-challengeStructured critical-thinking framework to pressure-test a decision before you commit.

Skille uruchamiają się na zwykłe polecenia — bez komend do zapamiętania. Po instalacji aktywują go prompty takie jak te (po angielsku):

  • Challenge this approach, are we over-engineering the caching layer?
  • Poke holes in my architecture plan before I commit to it
  • Play devil's advocate on my decision to store sessions only in Redis

Pełny raport →