Skip to content

alisru/Convergence-Test

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 

Repository files navigation

alisru/skills

Agent skills for structural analysis and judgement. Compatible with Claude.ai, Claude Code, and any agent runtime implementing the Agent Skills open standard.


Skills

convergence-test

Five-phase structural judgement protocol. Evaluates any idea, claim, policy, institution, or actor through the 7x7x7 Qqc tensor — a fractal interrogative structure that drills from surface claim to first principles.

Phases: Structural Scan → Vector Verification → Source Integrity → Forensic Stress Test → Verdict

Outputs: Hegemonic Grid coordinate (υ, ψ), R_net integrity score, ΔH hypocrisy gap, Fake Maximiser and Helxis detection, distortion node address.

Use when depth matters — locating exactly where a claim, institution, or actor breaks down.

Token cost: ~2,322 skill load · ~320 output per run · ~2,640 total per run

convergence-minimal

Single-pass binary scan across the same 7 planes. One note per plane (5–14 words), one summary line, no drilling.

Use for triage, orientation, or when the full protocol isn't warranted.

Token cost: ~471 skill load · ~210 output per run · ~680 total per run

Marginal cost of each additional subject in the same session: ~210 tokens (minimal) · ~320 tokens (full).


Sample results

Subject Skill Grid Zone Verdict
NATO Article 5 minimal (+0.8, +0.6) Greater Good ~
ESG investing minimal (−0.8, +0.4) Greatest Lie F
Universal Basic Income full (+0.9, +0.6) Greater Good ~
Mainstream media objectivity full (−0.9, +0.5) Greatest Lie F

Full sample outputs with token breakdowns: convergence-test-samples.md


Source markers

Both skills tag every plane claim with an epistemic marker:

  • 🌐[url] — searched this session, source linked inline
  • 🧠 — stable training knowledge
  • 🦕 — training knowledge, possibly outdated

Both skills search automatically when the subject involves recent events, live conflicts, current policy, or contested facts.


Installation

Drop the skill directory into your skills folder:

~/.config/claude/skills/convergence-test/
~/.config/claude/skills/convergence-minimal/

Or install via skillfish:

npm i skillfish
skillfish install alisru/convergence-test
skillfish install alisru/convergence-minimal

The framework

The 7 interrogative planes:

# Plane Interrogative Pass condition
Q1 Metaphysical WHO Stated beneficiary matches actual beneficiary
Q2 Possible WHAT Failure modes acknowledged, not suppressed
Q3 Physical WHERE Concrete, falsifiable physical prediction exists
Q4 Lyrical WHY Narrative honest; does not require obfuscation to hold
Q5 Logical HOW Causal chain holds under load
Q6 Historical CAUSE Origin acknowledged; history not erased to appear novel
Q7 Emotive EFFECT Emotional payload matches stated intent

The Hegemonic Grid:

  • υ axis (who benefits): +2 Everyone → 0 Neutral → −2 Only me
  • ψ axis (energy direction): +2 Creating systemic value → 0 Neutral → −2 Destroying/extracting

Zone anchors at ±1: Greater Good (+1,+1) · Greatest Lie (−1,+1) · Lesser Good (+1,−1) · Greater Evil (−1,−1)


License

MIT

About

Convergence testing to tell states ideals from actual actions

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors