LaunchKit · 2026
Back to Skills

core-refinery

Find the core that runs through everything — the ideas.

0
0 downloads
by @leegitw

Setup & Installation

openclaw skills install @leegitw/core-refinery

Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:

npx clawhub install core-refinery

What This Skill Does

Core Refinery takes 3 or more sources and identifies the principles that appear across all of them. It classifies ideas by how many sources they appear in: invariants (N≥3), domain-specific (N=2), or noise (N=1). Principles appearing in 3+ independent sources become Golden Master candidates.

The N-count threshold distinguishes genuine signal (3+ independent sources) from coincidence (2 sources), giving synthesis results a confidence level that simple summarization tools don't provide.

When to use it

  • Finding what multiple books on a topic all agree on
  • Consolidating recurring themes from several research papers
  • Identifying shared principles across team style guides or decision docs
  • Distilling common ground from multiple interview transcripts
  • Checking which ideas from different frameworks actually overlap

Example Workflow

Here's how your AI assistant might use this skill in practice.

INPUT

User asks: 'What's the core across these 4 extractions from books on writing?'

AGENT
  1. 1Collects all principles from the 4 provided sources
  2. 2Identifies threads — ideas that appear across multiple sources
  3. 3Tests whether the same meaning persists, not just the same words
  4. 4Classifies each principle as invariant (N≥3), domain-specific (N=2), or noise (N=1)
  5. 5Outputs Golden Master candidates with N-counts and confidence levels
OUTPUT

2 Golden Master candidates (N≥3), 5 domain-specific principles (N=2), and 8 filtered as noise — with compression ratio and suggested next steps