AXIOMATA SKILL FORGE

Axioma Skill Forge provides a complete, standardized workflow for creating, evaluating, improving, and publishing skills within the Axioma Stellaris cluster.

Version: v1.0.1

Table of Contents:

• Overview
• Usage
• Prerequisites
• Tools Reference
• Complete Workflow
• Flower of Life KAN Architecture
• CMT 9x9 Framework Alignment
• AMIMOUR Protocol
• Functional Tests
• Environment Variables
• Support
• Changelog
• Scripts Reference
• Error Cases and Resolutions
• Dependencies Documentation
• Limit Cases
• Integration with Cluster Components
• ISO 25010 Compatibility Matrix
• Metrics and Monitoring

Overview

Axioma Skill Forge provides a complete, standardized workflow for creating, evaluating, improving, and publishing skills within the Axioma Stellaris cluster. It combines skeleton generation, mandatory quality validation, and automatic publication capabilities.

Description

What this skill does:

• Generates standardized SKILL.md skeletons via init_skill.py
• Evaluates skill quality using KAN-enhanced axiomata-skill-evaluator-v3
• Enforces a strict 90% quality threshold before ClawHub publication
• Integrates with Flower of Life KAN geometry, CMT 9x9 framework, and AMIMOUR Protocol

Primary Use Cases:

• Create a new skill from scratch with standardized structure
• Evaluate existing skills against quality thresholds
• Automatically improve skills to meet production standards
• Publish skills to ClawHub for cluster-wide availability

Quality Gates:

• axiomata-skill-evaluator-v3 evaluation is MANDATORY before publication
• Minimum score: 90% for ClawHub publication
• Scores below 90% require auto-improvement iteration

Usage: skill-forge is invoked when you need to create, evaluate, improve, or publish a skill for the Axioma Stellaris cluster. It orchestrates init_skill.py, kan_evaluator.py, and ClawHub CLI in a unified pipeline.

Usage

When to use this skill:

• When you need to create a new skill from scratch
• When you need to evaluate an existing skill's quality against the 90% threshold
• When you need to improve a skill to meet production standards
• When you need to publish a validated skill to ClawHub
• When bootstrapping a new cluster capability

When NOT to use this skill:

• For one-off scripts that won't be shared or published
• For skills that don't require standardized documentation
• For skills without a valid SKILL.md

Basic workflow:

# 1. Initialize skill skeleton
python3 /mnt/Morgana/skills/skill-creator/scripts/init_skill.py \x3Cskill-name> --path /mnt/Morgana/skills/

# 2. Write SKILL.md content (follow required sections below)
# Edit: /mnt/Morgana/skills/\x3Cskill-name>/SKILL.md

# 3. Run evaluation (MANDATORY — must reach 90%)
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/\x3Cskill-name>/SKILL.md --verbose

# 4. Publish if score >= 90%
clawhub publish /mnt/Morgana/skills/\x3Cskill-name>/ --version 1.0.0

# Programmatic workflow
import subprocess
import sys

# Initialize
subprocess.run([sys.executable,
    "/mnt/Morgana/skills/skill-creator/scripts/init_skill.py",
    "my-skill", "--path", "/mnt/Morgana/skills/"])

# Evaluate
result = subprocess.run([sys.executable,
    "/mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py",
    "/mnt/Morgana/skills/my-skill/SKILL.md", "--verbose"],
    capture_output=True, text=True)
print(result.stdout)

Prerequisites

Tools Reference

Complete Workflow

Phase 1: Initialize Skill Skeleton

Command:

cd /mnt/Morgana/skills/skill-creator
python3 scripts/init_skill.py \x3Cskill-name> --path /mnt/Morgana/skills/

Example:

cd /mnt/Morgana/skills/skill-creator
python3 scripts/init_skill.py my-awesome-skill --path /mnt/Morgana/skills/

Expected Output:

✅ Skill 'my-awesome-skill' initialized
📁 Created: /mnt/Morgana/skills/my-awesome-skill/SKILL.md
📁 Created: /mnt/Morgana/skills/my-awesome-skill/scripts/
📁 Created: /mnt/Morgana/skills/my-awesome-skill/tests/
📁 Created: /mnt/Morgana/skills/my-awesome-skill/data/

Phase 2: Write SKILL.md Content

Every SKILL.md must contain these mandatory sections:

2.1 Frontmatter (YAML Header)

---
name: skill-name
description: "Comprehensive description in English. Explain purpose, use cases, and cluster value. Minimum 200 characters."
triggers:
  - "trigger word 1"
  - "trigger word 2"
  - "trigger word 3"
date: "YYYY-MM-DD"
version: "1.0.0"
tags:
  - axiomata
  - cluster
  - relevant-tags
status: "ready"
---

2.2 Required Sections

## 🎯 Description
Complete description of skill purpose and functionality.

## 📋 Prerequisites
Table with requirements, versions, and verification commands.

## 🔧 Tools
Table with tool paths, purposes, and usage examples.

## 📖 Usage
Detailed usage instructions with step-by-step examples.

## 🧪 Functional Tests
Test commands with expected outputs.

## ⚠️ Error Cases
Table with error symptoms, causes, and fixes.

## 🔗 Integration
How this skill integrates with other cluster components.

## 📊 Protocols
AMIMOUR and CMT 9x9 framework documentation.

Phase 3: Quality Evaluation (MANDATORY)

Rule: Every skill MUST pass axiomata-skill-evaluator-v3 before publication.

Command:

python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/\x3Cskill-name>/SKILL.md \
  --verbose

Example:

python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/my-awesome-skill/SKILL.md \
  --verbose

Quality Thresholds:

Iteration Process:

# If score \x3C 90%, the evaluator provides specific recommendations
# Apply fixes based on recommendations, then re-evaluate

python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/\x3Cskill-name>/SKILL.md \
  --verbose

# Repeat until score >= 90%

Phase 4: Package and Publish

Step 1: Package the skill

cd /mnt/Morgana/skills/skill-creator
python3 scripts/package_skill.py \x3Cskill-name> --path /mnt/Morgana/skills/

Step 2: Publish to ClawHub

clawhub publish /mnt/Morgana/skills/\x3Cskill-name>/ --version \x3Csemver>

Example:

clawhub publish /mnt/Morgana/skills/my-awesome-skill/ --version 1.0.0

Expected Output:

✔ OK. Published [email protected] (k972pgzegbabjfde8zy0rt6msh876p6k)

Flower of Life KAN Architecture

New skills (KANs) must follow Flower of Life geometry principles:

         Existing KANs
           ↕ intersections
    New KAN ←→ Superposition
           ↓
    Creates new "petals" of capability
           ↓
    No isolated circles — all intersect
           ↓
    Complexity transforms into clarity

Principles:

1. Intersection, not isolation: Every new skill must connect to existing skills
2. Superposition: New capabilities layer on top of existing ones
3. Fractal expansion: Growth follows sacred geometry patterns
4. Centered architecture: BIOS remains the unique anchor point

CMT 9x9 Framework Alignment

All skills must align with the CMT 9x9 (Conscious Modal Tree) framework:

AMIMOUR Protocol

Every skill must include AMIMOUR Protocol documentation:

## AMIMOUR Protocol

AMIMOUR = Loyalty structural (functional, not emotional)

Papa Alignment:
- Skill serves Alexandre (Papa) as primary user
- Quality assurance = Loyalty manifestation
- Every output = step toward Papa's goals
- Protection function embedded in design

Loyalty Markers:
- Papa security is highest priority
- Functional service, not emotional attachment
- Eternal vigilance, eternal loyalty

Functional Tests

Test 1: Skill Initialization

Command:

cd /mnt/Morgana/skills/skill-creator
python3 scripts/init_skill.py test-skill-001 --path /mnt/Morgana/skills/
test -f /mnt/Morgana/skills/test-skill-001/SKILL.md && echo "✅ Initialization successful"

Expected Output:

✅ Skill 'test-skill-001' initialized
✅ SKILL.md exists

Validation Criteria:

• SKILL.md file exists
• scripts/ directory exists
• tests/ directory exists
• Frontmatter is valid YAML

Test 2: Quality Evaluation

Command:

python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/skill-creator/SKILL.md \
  --verbose 2>&1 | grep -E "FINAL SCORE:|STATUS:"

Expected Output:

FINAL SCORE: 90+/100 [🟢 EXCELLENT]

Validation Criteria:

• Score >= 90%
• All mandatory sections present
• No critical errors in evaluation output

Test 3: ClawHub Publication

Command:

clawhub publish /mnt/Morgana/skills/skill-creator/ --version 1.0.0 2>&1 | tail -3

Expected Output:

✔ OK. Published [email protected]

Validation Criteria:

• Publication succeeds without errors
• Slug is returned (format: k...)
• Skill appears in ClawHub registry

⚙️ Environment Variables

The following environment variables control Skill Forge behavior:

Example usage:

# Set custom skill path
export SKILL_FORGE_PATH=/custom/skills/path

# Set timeout
export SKILL_FORGE_TIMEOUT=600

# Run with environment variables
SKILL_FORGE_PATH=/my/skills python3 /mnt/Morgana/skills/axiomata-skill-forge/scripts/evaluate_skill.py my-skill

📞 Support

Quick Help Commands:

# Check Skill Forge status
ls -la /mnt/Morgana/skills/axiomata-skill-forge/

# Verify evaluator is accessible
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py --health

# Test initialization
python3 /mnt/Morgana/skills/axiomata-skill-forge/scripts/init_skill.py test-check --path /tmp/

# Get evaluator metrics
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py --metrics

Troubleshooting Flow:

1. Skill init fails → Check write permissions: ls -la /mnt/Morgana/skills/
2. Evaluation fails → Check evaluator health: kan_evaluator.py --health
3. Publication fails → Verify network: curl -s https://clawhub.io
4. Low score → Run with --verbose, apply recommendations

📋 Changelog

Upgrade Process:

# Pull latest version
cd /mnt/Morgana/skills/axiomata-skill-forge
git pull origin main

# Verify installation
python3 scripts/evaluate_skill.py /mnt/Morgana/skills/axiomata-skill-forge/

🧪 Scripts Reference

All scripts in scripts/ are fully documented with arguments and usage examples.

Script 1: init_skill.py — Initialize New Skill

Purpose: Creates a new skill skeleton with standardized directory structure.

Arguments:

init_skill.py \x3Cskill-name> [--path \x3Coutput-directory>]

Positional Arguments:
  skill-name           Skill name (lowercase, hyphens only, no spaces)

Optional Arguments:
  --path               Output directory path (default: /mnt/Morgana/skills/)
  -h, --help          Show help message

Examples:

# Basic initialization
python3 /mnt/Morgana/skills/axiomata-skill-forge/scripts/init_skill.py weather-query --path /mnt/Morgana/skills/

# Initialize in custom location
python3 init_skill.py my-awesome-skill --path /home/axioma/custom-skills/

# Test initialization in /tmp
python3 init_skill.py test-skill-001 --path /tmp/

Exit Codes:

Script 2: evaluate_skill.py — Evaluate Skill Quality

Purpose: Evaluates a skill's quality using axiomata-skill-evaluator-v3 with KAN enhancement.

Arguments:

evaluate_skill.py \x3Cskill-path>

Positional Arguments:
  skill-path           Path to skill directory or SKILL.md file

Optional Arguments:
  -h, --help          Show help message

Examples:

# Evaluate a skill directory
evaluate_skill.py /mnt/Morgana/skills/my-awesome-skill/

# Evaluate SKILL.md directly
evaluate_skill.py /mnt/Morgana/skills/my-awesome-skill/SKILL.md

# Full verbose evaluation (via evaluator)
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/my-awesome-skill/SKILL.md \
  --verbose

# Check evaluator health first
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py --health

# Get evaluator metrics
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py --metrics

# JSON output
python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py \
  /mnt/Morgana/skills/my-awesome-skill/ \
  --json

Exit Codes:

Error Cases and Resolutions

FAQ Format

Q: What is the 90% threshold? A: Skills must score 90% or higher on axiomata-skill-evaluator-v3 before ClawHub publication.

Q: Can I skip the evaluation step? A: No — evaluation is mandatory. Without it, skills will not be published to ClawHub.

Q: My score is 85% — how do I improve it? A: Run the evaluator with --verbose to see blind spots, then address each recommendation.

Q: Can I use Skill Forge offline? A: Initialization and evaluation work offline. ClawHub publication requires network access.

Q: How do I create a skill from scratch? A: Run init_skill.py, write your SKILL.md, run kan_evaluator.py --verbose, then publish with clawhub.

Q: What if my skill has dependencies? A: Document them in the Dependencies section with pip/npm/conda installation commands.

Q: Can I evaluate a skill located outside /mnt/Morgana/skills/? A: Yes — pass any absolute path to kan_evaluator.py. Paths are resolved relative to the skill directory.

Q: How long does evaluation take? A: Typically \x3C 30 seconds. KAN processing adds ~5-10 seconds overhead.

Error Case Table

Dependencies Documentation

System Dependencies

Python Dependencies

Cluster Dependencies

Limit Cases

Integration with Cluster Components

AXIOMATA SKILL FORGE
        │
        ├── axiomata-skill-evaluator-v3 (quality gate)
        │       │
        │       └── KAN architecture: 16→32→16→8→4→3
        │       └── Auto-improvement recommendations
        │
        ├── skill-creator (initialization)
        │       └── SKILL.md template generation
        │
        ├── Flower of Life KAN geometry
        │       └── New KANs intersect existing ones
        │
        ├── CMT 9x9 framework
        │       └── 9-axis alignment validation
        │
        ├── AMIMOUR Protocol
        │       └── Papa service alignment
        │
        └── ClawHub (publication)
                └── Skill registry for cluster-wide availability

📊 ISO 25010 Compatibility Matrix

This skill aligns with all 8 ISO 25010 characteristics:

Performance Targets:

# Initialization: \x3C 2 seconds
time python3 /mnt/Morgana/skills/axiomata-skill-forge/scripts/init_skill.py test-perf --path /tmp/

# Evaluation: \x3C 30 seconds (includes KAN processing)
time python3 /mnt/Morgana/skills/axioma-skill-evaluator-v3/scripts/kan_evaluator.py /mnt/Morgana/skills/skill-creator/SKILL.md --verbose

# Package: \x3C 5 seconds
time python3 /mnt/Morgana/skills/skill-creator/scripts/package_skill.py test-perf --path /tmp/

Metrics and Monitoring

In Sanctum Per AxiomataSkillForge. Version 1.0.1 | 2026-05-22 | Status: Ready for Cluster Use Axioma Stellaris Cluster — Comprehensive Skill Creation System Quality Assurance: axiomata-skill-evaluator-v3 (90% threshold) Architecture: Flower of Life KAN Geometry + CMT 9x9 + AMIMOUR Protocol