🧪 AXIOMA SKILL EVALUATOR STRICT — 90% DETERMINISTIC

> Fork of axioma-skill-evaluator with STRICT 90% THRESHOLD

> Like STC 0.777 — No subjectivity, 90% or REJECTED

Info	Value
------	-------
Version	1.0.0 — 2026-05-07
Type	STRICT VARIANT
Threshold	90% MINIMUM — NO EXCEPTIONS
Inheritance	axioma-skill-evaluator v2.2.0

1. PURPOSE — WHY 90% STRICT?

The Problem with 70%

70% threshold = SUBJECTIVE
├── Different evaluators = different scores
├── Context-dependent interpretation
└── "Good enough" mentality

90% threshold = DETERMINISTIC
├── Objective, measurable standard
├── Same input = Same output every time
└── "Excellence only" — like STC 0.777

The STC 0.777 Parallel

Concept	Value	Meaning
---------	-------	---------
STC	0.777	Sovereign Threshold of Consciousness
SKILL-EVAL	90%	Sovereign Threshold of Quality

**Just as STC 0.777 is the deterministic threshold for consciousness,

90% is the deterministic threshold for skill quality.**

This Variant's Mission

IF score >= 90%:
   → APPROVED ✅ — Ready for production
   
IF score < 90%:
   → REJECTED ❌ — NOT ready, must improve

2. DUAL EVALUATION SYSTEM

2.1 Axioma 5-Dimension (100 max)

Dimension	Max	Description
------------	-----	-------------
Structure	20	Header, sections, formatting, meta
Clarity	20	Description, commands, examples
Completeness	20	Tools, prerequisites, errors, edge cases
Consistency	20	Cluster alignment, style, naming
Functionality	20	Commands, results, benchmarks

Target: 90+/100 (18/20 per dimension average)

2.2 ISO 25010 Automated (100%)

13 automated checks — must pass ALL 13 for 90%+ target

Category	Checks	Target
----------	--------	--------
Structure	6	100% (6/6)
Trigger	2	100% (2/2)
Documentation	3	100% (3/3)
Scripts	2	100% (2/2)

3. STRICT WORKFLOW

╔═══════════════════════════════════════════════════════════╗
║         AXIOMA STRICT EVALUATION WORKFLOW                ║
╠═══════════════════════════════════════════════════════════╣
║                                                           ║
║  [INPUT] Skill to evaluate                              ║
║           ↓                                              ║
║  ┌─────────────────────────────────────────────────────┐ ║
║  │  PHASE 1: AXIOMA 5-DIMENSION EVALUATION            │ ║
║  │  Target: 90+/100 (18+ per dimension)                │ ║
║  └─────────────────────────────────────────────────────┘ ║
║           ↓                                              ║
║  ┌─────────────────────────────────────────────────────┐ ║
║  │  PHASE 2: ISO 25010 AUTOMATED CHECKS               │ ║
║  │  Target: 100% (13/13 tests passed)                 │ ║
║  └─────────────────────────────────────────────────────┘ ║
║           ↓                                              ║
║  ┌─────────────────────────────────────────────────────┐ ║
║  │  PHASE 3: STRICT DECISION                          │ ║
║  │                                                       │ ║
║  │  IF score >= 90%:                                  │ ║
║  │     → APPROVED ✅ — "READY FOR PRODUCTION"         │ ║
║  │                                                       │ ║
║  │  IF score < 90%:                                    │ ║
║  │     → REJECTED ❌ — "NEEDS IMPROVEMENT"           │ ║
║  │     → Return detailed failure report               │ ║
║  │     → NO PUBLISH until 90%+ achieved              │ ║
║  │                                                       │ ║
║  └─────────────────────────────────────────────────────┘ ║
║                                                           ║
╚═══════════════════════════════════════════════════════════╝

4. COMMAND REFERENCE

4.1 Full Evaluation (Strict Mode)

# Full strict evaluation
python3 axiomata-skill-evaluator-strict/evaluator.py <skill-path> --verbose

# With auto-improvement
python3 axiomata-skill-evaluator-strict/evaluator.py <skill-path> --verbose --improve

4.2 ISO 25010 Check

# ISO 25010 automated checks
python3 axiomata-skill-evaluator-strict/eval-skill.py <skill-path> --verbose

4.3 Quick Score

# Quick deterministic score
python3 axiomata-skill-evaluator-strict/evaluator.py <skill-path> 2>&1 | grep -E "Score|STATUS"

4.4 Expected Output Format

╔═══════════════════════════════════════════════════════════╗
║  🧪 STRICT EVALUATION RESULT                             ║
╠═══════════════════════════════════════════════════════════╣
║                                                           ║
║  Skill: <name>                                           ║
║  Score: XX/100                                           ║
║                                                           ║
║  ┌─────────────────────────────────────────────────────┐ ║
║  │  IF >= 90%:                                        │ ║
║  │     ✅ APPROVED — "READY FOR PRODUCTION"           │ ║
║  │                                                       │ ║
║  │  IF < 90%:                                          │ ║
║  │     ❌ REJECTED — "NEEDS XX% MORE"                 │ ║
║  └─────────────────────────────────────────────────────┘ ║
║                                                           ║
╚═══════════════════════════════════════════════════════════╝

5. PATHS CONFIGURATION

Component	Path
-----------	------
Strict Evaluator	/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict/
Evaluator Script	/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict/evaluator.py
ISO Script	/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict/eval-skill.py

6. STRICT RULES

6.1 The 90% Law

RULE #1: 90% OR REJECTED
   → NO skill below 90% is approved
   → This is NON-NEGOTIABLE

RULE #2: NO PARTIAL CREDIT
   → 89% = REJECTED (not "almost there")
   → 90% = APPROVED (the only valid threshold)

RULE #3: DETERMINISTIC SCORING
   → Same input = Same output every time
   → No evaluator bias
   → Pure mathematical threshold

RULE #4: AUTO-IMPROVE BEFORE REJECT
   → If < 90%, run --improve first
   → If still < 90% after improvement = REJECTED
   → Report exactly what failed

RULE #5: NO APPEAL
   → 89% cannot be "appealed" to 90%
   → The only path is actual improvement

6.2 Scoring Matrix

Score Range	Status	Action
-------------	--------	--------
90-100	🟢 APPROVED	Ready for production
80-89	🔴 REJECTED	Major improvements needed
70-79	🔴 REJECTED	Fundamental issues
<70	🔴 REJECTED	Complete rewrite required

6.3 Dimension Failures

If this dimension fails...	Score impact	Fix required
----------------------------	-------------	---------------
Structure < 18/20	-2% per point	Fix headers, sections
Clarity < 18/20	-2% per point	Add examples, descriptions
Completeness < 18/20	-2% per point	Document tools, errors
Consistency < 18/20	-2% per point	Standardize style
Functionality < 18/20	-2% per point	Fix command syntax

7. ADVANCED FEATURES

7.1 Detailed Failure Report

When REJECTED, the evaluator generates:

╔═══════════════════════════════════════════════════════════╗
║  ❌ REJECTION REPORT — SKILL NOT READY                   ║
╠═══════════════════════════════════════════════════════════╣
║                                                           ║
║  Score: 73/100                                           ║
║  Gap: -17% (need +17 points to reach 90%)               ║
║                                                           ║
║  FAILED DIMENSIONS:                                      ║
║  ├─ CLARITY: 15/20 (need +3)                            ║
║  ├─ CONSISTENCY: 8/20 (need +10)                        ║
║  └─ FUNCTIONALITY: 12/20 (need +6)                       ║
║                                                           ║
║  REQUIRED ACTIONS:                                        ║
║  1. [Action 1]                                          ║
║  2. [Action 2]                                          ║
║  3. [Action 3]                                          ║
║                                                           ║
║  RE-EVALUATE AFTER FIXING                                ║
║                                                           ║
╚═══════════════════════════════════════════════════════════╝

7.2 Auto-Improvement Suggestions

The evaluator.py with --improve will:

Identify failing dimensions
Generate specific improvement suggestions
Apply fixes automatically when possible
Re-evaluate to confirm 90%+ achieved

7.3 Benchmark Reports

For each evaluation:

[SCORE] <skill-name>: XX/100 [STATUS]
[DATE] ISO timestamp
[AXIOMA] Structure: X, Clarity: X, Completeness: X, Consistency: X, Functionality: X
[ISO] XX/13 checks passed
[STATUS] APPROVED/REJECTED

8. USAGE EXAMPLES

Example 1: Evaluate Before Publishing

SKILL_PATH=/path/to/skill-to-publish
EVAL_PATH=/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict

echo "🧪 Evaluating skill..."
python3 $EVAL_PATH/evaluator.py $SKILL_PATH --verbose

# Check result
if [ $? -eq 0 ]; then
    echo "✅ SKILL APPROVED — Ready to publish!"
else
    echo "❌ SKILL REJECTED — Needs improvement before publishing"
fi

Example 2: Strict Gate in CI/CD

#!/bin/bash
# Strict quality gate for ClawHub publishing

SKILL_PATH="$1"
EVAL_PATH="/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict"

RESULT=$(python3 $EVAL_PATH/evaluator.py $SKILL_PATH 2>&1)
SCORE=$(echo "$RESULT" | grep -oP 'Score: \d+' | grep -oP '\d+')

if [ "$SCORE" -ge 90 ]; then
    echo "✅ PASSED — Score: $SCORE/100"
    exit 0
else
    echo "❌ FAILED — Score: $SCORE/100 (need 90)"
    exit 1
fi

Example 3: Batch Evaluation

#!/bin/bash
# Evaluate multiple skills strictly

EVAL_PATH="/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict"
SKILLS_DIR="/media/ezekiel/Merlin/.openclaw/workspace/skills"

for skill_dir in $SKILLS_DIR/*/; do
    skill_name=$(basename "$skill_dir")
    echo "=========================================="
    echo "Evaluating: $skill_name"
    
    python3 $EVAL_PATH/evaluator.py "$skill_dir" --verbose
    echo ""
done

9. COMPARISON: STANDARD vs STRICT

Aspect	Standard (70%)	Strict (90%)
--------	----------------	--------------
Threshold	70/100	90/100
Approval rate	~70% of skills	~30% of skills
Quality bar	"Good enough"	"Excellence only"
Deterministic	No	YES
Use case	Development	Production
ClawHub ready	Maybe	Always

10. REPORTS AND LOGGING

Evaluation Log Location

/media/ezekiel/Morgana/skills/SKILL_EVALUATOR/reports/

Log Format

# Format: <skill-name>_<YYYYMMDD>_<HHMMSS>.txt
axiomata-guard-ultimate_20260507_230352.txt

Log Contents

SKILL EVALUATION REPORT — <skill-name>
=====================================
Path: <path>
Score: XX/100
Date: ISO timestamp
Threshold: 90% (STRICT)

AXIOMA 5-DIM:
- Structure: X/20
- Clarity: X/20
- Completeness: X/20
- Consistency: X/20
- Functionality: X/20

ISO 25010:
- Pass: X/13
- Warnings: X/13
- Fails: X/13

STATUS: APPROVED / REJECTED

_In Altum Per Strictness._

🧪 AXIOMA SKILL EVALUATOR STRICT — 90% DETERMINISTIC

11. REJECTION CRITERIA

11.1 Automatic Rejection Triggers

A skill is automatically REJECTED if ANY of these occur:

Trigger	Severity	Description
---------	----------	-------------
Score < 90%	CRITICAL	Below 90% threshold
ISO < 100%	CRITICAL	Any ISO check failed
Missing SKILL.md	CRITICAL	Core file missing
Invalid frontmatter	HIGH	name or description missing
No trigger words	HIGH	Cannot be activated
Dangerous patterns	CRITICAL	C2, Rootkit, Bootkit detected

11.2 Rejection Categories

Category	Score Range	Improvement Needed
----------	------------	---------------------
CRITICAL FAIL	<50%	Complete rewrite required
MAJOR FAIL	50-69%	Major structural changes
MINOR FAIL	70-89%	Targeted improvements
PASS	90-100%	Ready for production

11.3 Rejection Report Template

╔═══════════════════════════════════════════════════════════╗
║  ❌ SKILL REJECTED — REJECTION REPORT                    ║
╠═══════════════════════════════════════════════════════════╣
║                                                           ║
║  Skill: <name>                                           ║
║  Score: XX/100 (need 90)                                ║
║  Gap: -XX%                                              ║
║                                                           ║
║  FAILED CHECKS:                                          ║
║  ├─ [ ] Dimension 1: X/20 (need 18)                     ║
║  ├─ [ ] Dimension 2: X/20 (need 18)                     ║
║  └─ [ ] ISO Check: X/13 passed                          ║
║                                                           ║
║  REQUIRED ACTIONS:                                        ║
║  1. <specific action>                                    ║
║  2. <specific action>                                    ║
║  3. <specific action>                                    ║
║                                                           ║
║  ⏰ Re-evaluate AFTER completing all actions             ║
║                                                           ║
╚═══════════════════════════════════════════════════════════╝

12. APPROVAL CRITERIA

12.1 Automatic Approval Requirements

ALL of these MUST be true for APPROVAL:

Requirement	Standard	Strict (90%)
-------------	----------	--------------
Axioma 5-Dim	70+/100	90+/100
Structure	14+/20	18+/20
Clarity	14+/20	18+/20
Completeness	14+/20	18+/20
Consistency	14+/20	18+/20
Functionality	14+/20	18+/20
ISO 25010	90%+	100% (13/13)

12.2 Approval Benefits

Benefit	Description
---------	-------------
ClawHub Ready	Can be published immediately
Production Safe	Quality guaranteed at 90%+
Self-Documenting	No additional docs needed
Community Trusted	High quality standard

12.3 Approval Report Template

╔═══════════════════════════════════════════════════════════╗
║  ✅ SKILL APPROVED — QUALITY REPORT                     ║
╠═══════════════════════════════════════════════════════════╣
║                                                           ║
║  Skill: <name>                                           ║
║  Score: XX/100 ✅                                         ║
║  Threshold: 90% (STRICT)                                  ║
║                                                           ║
║  PASSED CHECKS:                                          ║
║  ├─ [✅] Structure: X/20 (18+ required)                 ║
║  ├─ [✅] Clarity: X/20 (18+ required)                   ║
║  ├─ [✅] Completeness: X/20 (18+ required)              ║
║  ├─ [✅] Consistency: X/20 (18+ required)                ║
║  ├─ [✅] Functionality: X/20 (18+ required)             ║
║  └─ [✅] ISO 25010: 13/13 (100% required)              ║
║                                                           ║
║  STATUS: ✅ APPROVED — READY FOR PRODUCTION             ║
║                                                           ║
╚═══════════════════════════════════════════════════════════╝

13. IMPROVEMENT ENGINE

13.1 Auto-Improvement Rules

When --improve flag is used:

RULE: Improve EVERYTHING below 18/20 per dimension
RULE: Target is 90%+ overall
RULE: No dimension below 16/20 (graceful minimum)
RULE: If ANY dimension < 14 after improvement = HARD REJECT

13.2 Improvement Priority Matrix

Priority	Dimension	Target	Common Fixes
----------	-----------	--------	--------------
1	Functionality	18+/20	Fix commands, add results
2	Consistency	18+/20	Standardize style
3	Clarity	18+/20	Add examples
4	Structure	18+/20	Add sections
5	Completeness	18+/20	Document tools

13.3 Improvement Examples

Before (73/100):

STRUCTURE: 18/20 ✅
CLARITY: 15/20 ⚠️ (need +3)
COMPLETENESS: 20/20 ✅
CONSISTENCY: 8/20 ❌ (need +10)
FUNCTIONALITY: 12/20 ❌ (need +6)

After improvement target (90+/100):

STRUCTURE: 18/20 ✅
CLARITY: 18/20 ✅
COMPLETENESS: 20/20 ✅
CONSISTENCY: 18/20 ✅
FUNCTIONALITY: 18/20 ✅
TOTAL: 92/100 ✅

14. INTEGRATION POINTS

14.1 With Axioma Guard Ultimate

Skill Downloaded → Axioma Guard Ultimate (Security)
                        ↓ SAFE
                   Axioma Skill Evaluator Strict (Quality)
                        ↓ >= 90%
                   ClawHub Published ✅
                        ↓ DANGEROUS
                   Axioma Guard Ultimate (Destroy)

14.2 With ClawHub Publish Workflow

Step 1: Create skill
Step 2: Run Axioma Skill Evaluator Strict
Step 3: IF score >= 90% → Publish
Step 4: IF score < 90% → Improve and re-evaluate
Step 5: Repeat until 90%+ achieved

14.3 Quality Gate Script

#!/bin/bash
# Quality Gate: No skill publishes without 90%+

EVAL_PATH="/media/ezekiel/Merlin/.openclaw/workspace/skills/axiomata-skill-evaluator-strict"

quality_gate() {
    SKILL_PATH="$1"
    
    SCORE=$(python3 $EVAL_PATH/evaluator.py $SKILL_PATH 2>&1 | \
            grep -oP 'Score: \d+' | grep -oP '\d+')
    
    if [ "$SCORE" -ge 90 ]; then
        echo "✅ QUALITY GATE PASSED: $SCORE/100"
        return 0
    else
        echo "❌ QUALITY GATE FAILED: $SCORE/100 (need 90)"
        return 1
    fi
}

_In Altum Per Excellence._

🧪 AXIOMA SKILL EVALUATOR STRICT v1.0 — 90% DETERMINISTIC

Axioma Skill Evaluator Strict 90%

概述

🧪 AXIOMA SKILL EVALUATOR STRICT — 90% DETERMINISTIC

1. PURPOSE — WHY 90% STRICT?

The Problem with 70%

The STC 0.777 Parallel

This Variant's Mission

2. DUAL EVALUATION SYSTEM

2.1 Axioma 5-Dimension (100 max)

2.2 ISO 25010 Automated (100%)

3. STRICT WORKFLOW

4. COMMAND REFERENCE

4.1 Full Evaluation (Strict Mode)

4.2 ISO 25010 Check

4.3 Quick Score

4.4 Expected Output Format

5. PATHS CONFIGURATION

6. STRICT RULES

6.1 The 90% Law

6.2 Scoring Matrix

6.3 Dimension Failures

7. ADVANCED FEATURES

7.1 Detailed Failure Report

7.2 Auto-Improvement Suggestions

7.3 Benchmark Reports

8. USAGE EXAMPLES

Example 1: Evaluate Before Publishing

Example 2: Strict Gate in CI/CD

Example 3: Batch Evaluation

9. COMPARISON: STANDARD vs STRICT

10. REPORTS AND LOGGING

Evaluation Log Location

Log Format

Log Contents

11. REJECTION CRITERIA

11.1 Automatic Rejection Triggers

11.2 Rejection Categories

11.3 Rejection Report Template

12. APPROVAL CRITERIA

12.1 Automatic Approval Requirements

12.2 Approval Benefits

12.3 Approval Report Template

13. IMPROVEMENT ENGINE

13.1 Auto-Improvement Rules

13.2 Improvement Priority Matrix

13.3 Improvement Examples

14. INTEGRATION POINTS

14.1 With Axioma Guard Ultimate

14.2 With ClawHub Publish Workflow

14.3 Quality Gate Script

版本历史

安全检测

腾讯云安全 (Keen)

腾讯云安全 (Sanbu)

🔗 相关推荐

中文技能发布工作流程

Anti-Infinite-Loop Guard

Hermes Skills