GlyphRunner System
488 lines
10 KiB
Markdown
Executable File
488 lines
10 KiB
Markdown
Executable File
# Glyph Compression Executor — Technical Documentation
|
||
|
||
## Architecture Overview
|
||
|
||
### Dual-Layer Symbolic System
|
||
|
||
SuperDave 2125 implements a **dual-layer architecture**:
|
||
|
||
#### Computational Layer
|
||
- **Purpose**: Execute Python code through compressed binary format
|
||
- **Components**:
|
||
- GSZ3 compression (zlib + SHA256 checksum)
|
||
- XIC binary format (XIC header + JSON manifest + compressed payload)
|
||
- LAIN 8-lane cognition engine
|
||
- Segment runtime executor
|
||
|
||
#### Symbolic Layer
|
||
- **Purpose**: Analyze code through 600 specialized glyphs with 152 superpowers
|
||
- **Components**:
|
||
- LedoGlyph600 registry (600 glyphs)
|
||
- Superpower registry (152 superpowers)
|
||
- Multi-glyph resonance calculation
|
||
- Glyph activation from intent
|
||
|
||
### Data Flow
|
||
|
||
```
|
||
Python Source → GSZ3 Compress → XIC Pack → LAIN Cognition → Execution Result
|
||
```
|
||
|
||
### Compression Pipeline
|
||
|
||
1. **Segmentation**: Split code into logical segments
|
||
2. **Compression**: GSZ3 format (zlib level 9 + SHA256[:3] checksum)
|
||
3. **Packing**: XIC binary format with JSON manifest
|
||
4. **Execution**: Decompress → Execute through LAIN → Return fused symbol
|
||
|
||
---
|
||
|
||
## 600 Glyphs System
|
||
|
||
### G001 (Ledo/Aether Node) - The Root Glyph
|
||
|
||
**Unique Properties**:
|
||
- **152 superpowers** (ALL available)
|
||
- **Specialized Type**: `aether_node`
|
||
- **Power Boost**: 387.95x (38,695% effectiveness increase)
|
||
- **VRAM Budget**: 7.5GB (maximum for GTX 1080)
|
||
- **Priority**: 10.0 (maximum)
|
||
- **Constraints**: None (primordial authority)
|
||
- **Enhancements**: `universal_override`, `primordial_resonance`, `system_root_access`
|
||
|
||
**Purpose**: G001 is the **primordial root glyph** that holds all system authority. It cannot be replicated by any other glyph.
|
||
|
||
### Other Glyphs (G002-G600)
|
||
|
||
**Superpower Limits**:
|
||
- **Min**: 9 superpowers
|
||
- **Max**: 22 superpowers
|
||
- **Most Common**: 15 superpowers (269 glyphs)
|
||
|
||
**Distribution**:
|
||
```
|
||
9-10: 7 glyphs
|
||
11-12: 54 glyphs
|
||
13-15: 485 glyphs
|
||
16-22: 54 glyphs
|
||
152: 1 glyph (G001 only)
|
||
```
|
||
|
||
### Glyph Categories
|
||
|
||
| Category | Count | Purpose |
|
||
|----------|-------|---------|
|
||
| neural | 75 | Core cognition |
|
||
| communication | 72 | Data transfer |
|
||
| defense | 68 | Security |
|
||
| energy | 65 | Power management |
|
||
| life-support | 62 | System stability |
|
||
| navigation | 58 | Path finding |
|
||
| propulsion | 55 | Movement control |
|
||
| research | 55 | Discovery |
|
||
|
||
---
|
||
|
||
## 152 Superpowers
|
||
|
||
### Superpower Bands
|
||
|
||
| Band | Range | Purpose |
|
||
|------|-------|---------|
|
||
| A | 1-15 | Foundational operations |
|
||
| B | 16-45 | Advanced processing |
|
||
| C | 46-76 | Specialized functions |
|
||
| D | 77-152 | Advanced capabilities |
|
||
|
||
### Boost Calculation
|
||
|
||
```python
|
||
power_boost = 1.0 + Σ(boost_percent) / 100.0
|
||
```
|
||
|
||
**Example**:
|
||
- G001 with 152 superpowers: **387.95x**
|
||
- G002 with 18 superpowers: **14.50x**
|
||
- G050 with 15 superpowers: **8.25x**
|
||
|
||
### Top Superpowers
|
||
|
||
| ID | Name | Boost | Band |
|
||
|----|------|-------|------|
|
||
| 1 | DNA Supercoiling Access | +65% | A |
|
||
| 77 | MOF Fluidic Ion Transistor | +250% | D |
|
||
| 100 | Superheavy Element Synthesis | +450% | D |
|
||
| 152 | Neuralink-Style Brain-Computer Interface | +480% | D |
|
||
|
||
---
|
||
|
||
## GSZ3 Compression
|
||
|
||
### Format Specification
|
||
|
||
```
|
||
Header (12 bytes):
|
||
[0-3] Magic: "GSZ3" (0x47535A33)
|
||
[4] Version: 1
|
||
[5-8] Payload Length (uint32, big-endian)
|
||
[9-11] Checksum: SHA256(payload)[:3]
|
||
|
||
Payload:
|
||
zlib level 9 compressed data
|
||
```
|
||
|
||
### Compression Algorithm
|
||
|
||
1. UTF-8 encode text
|
||
2. zlib compress (level 9)
|
||
3. SHA256 hash compressed data
|
||
4. Take first 3 bytes as checksum
|
||
5. Concatenate: Magic + Version + Length + Checksum + Compressed Data
|
||
|
||
### Decompression Algorithm
|
||
|
||
1. Verify magic number
|
||
2. Read version
|
||
3. Read payload length
|
||
4. Verify checksum
|
||
5. zlib decompress
|
||
6. UTF-8 decode
|
||
|
||
---
|
||
|
||
## LAIN 8-Lane Symbolic Cognition
|
||
|
||
### Lane Assignment Algorithm
|
||
|
||
Lanes are assigned based on **segment content analysis**:
|
||
|
||
```python
|
||
def _infer_lane_from_content(content):
|
||
if has_control_flow: return 0 # if, for, while, return, try, except, with
|
||
elif has_comments: return 3 # #, //, /*, */
|
||
elif has_hints: return 4 # hint, note, todo, fixme, warning, danger
|
||
elif has_metadata: return 3 # <glyph:, metadata, tag, annotation
|
||
elif has_execution_hints: return 4 # rm -rf, del, os.system, subprocess
|
||
elif has_template: return 5 # step, todo:, placeholder, fill-in
|
||
elif has_contributor: return 6 # author, contributor, copyright, @
|
||
elif has_epoch: return 7 # epoch, timestamp, date, time, version
|
||
else: return 1 # default semantic flow
|
||
```
|
||
|
||
### Lane Processing
|
||
|
||
Each lane processes segments with specialized handlers:
|
||
|
||
| Lane | Processor | Output |
|
||
|------|-----------|--------|
|
||
| 0 | structural_logic | Control flow analysis |
|
||
| 1 | semantic_flow | Core meaning extraction |
|
||
| 2 | compression_residue | Artifact detection |
|
||
| 3 | symbolic_metadata | Tag/annotation analysis |
|
||
| 4 | execution_hints | Safety analysis |
|
||
| 5 | predictive_scaffolding | Pattern prediction |
|
||
| 6 | contributor_imprint | Author style detection |
|
||
| 7 | epoch_resonance | Temporal context |
|
||
|
||
---
|
||
|
||
## Multi-Glyph Resonance
|
||
|
||
### Calculation Formula
|
||
|
||
For each glyph, compute 5-dimensional metrics:
|
||
|
||
```python
|
||
weight = (glyph_score / 335) * 0.7 + (activation_score / 100) * 0.3
|
||
lineage_score = inheritance_weight
|
||
contributor_score = connectivity / 100
|
||
frequency_score = sqrt(P² + R² + A² + W²) / 200
|
||
grammar_score = stability / 100
|
||
```
|
||
|
||
### Global Resonance
|
||
|
||
```python
|
||
global_resonance = Σ(weight) / count
|
||
```
|
||
|
||
---
|
||
|
||
## Superpower Assignment Algorithm
|
||
|
||
### Power Count Formula
|
||
|
||
```python
|
||
power_count = 5 + int((avg_metric / 100) * 20)
|
||
```
|
||
|
||
Where `avg_metric = (power + complexity + resonance + stability + connectivity + affinity) / 6`
|
||
|
||
### Band Eligibility
|
||
|
||
| Tier | Bands | Rule |
|
||
|------|-------|------|
|
||
| G001 | A, B, C, D | Aether node (all bands) |
|
||
| G002-G150 | A, B | Tier 1-15 |
|
||
| G151-G300 | B, C | Tier 16-30 |
|
||
| G301-G450 | C, D | Tier 31-45 |
|
||
| G451-G600 | D, C | Tier 46-60 |
|
||
|
||
### Superpower Scoring
|
||
|
||
```python
|
||
score = 0.45 × metrics + 0.35 × type_bias + 0.15 × boost% + 0.05 × hash
|
||
```
|
||
|
||
Where:
|
||
- `metrics` = average of glyph metrics (0-100)
|
||
- `type_bias` = 100 if preferred, 25 if not
|
||
- `boost%` = superpower boost percentage
|
||
- `hash` = deterministic variety (MD5 of glyph_id + superpower_id)
|
||
|
||
---
|
||
|
||
## File Format Specifications
|
||
|
||
### .gx Binary Format
|
||
|
||
```
|
||
Header (8 bytes):
|
||
[0-2] Magic: "XIC" (0x584943)
|
||
[3] Version: 1
|
||
[4-7] Manifest Length (uint32, big-endian)
|
||
|
||
Manifest (variable length):
|
||
JSON with keys:
|
||
- magic: "GXIC1"
|
||
- version: 1
|
||
- source_file: str
|
||
- source_type: str
|
||
- version_str: str
|
||
- contributor: str
|
||
- timestamp: ISO 8601
|
||
- codex_lineage: {
|
||
segments: [{
|
||
id: str,
|
||
start: int,
|
||
end: int,
|
||
start_byte: int,
|
||
end_byte: int
|
||
}]
|
||
}
|
||
|
||
Payload:
|
||
GSZ3 compressed data
|
||
```
|
||
|
||
### JSON Manifest Format
|
||
|
||
```json
|
||
{
|
||
"magic": "GXIC1",
|
||
"version": 1,
|
||
"source_file": "test.py",
|
||
"source_type": ".py",
|
||
"version_str": "1.0.0",
|
||
"contributor": "GlyphRunner",
|
||
"timestamp": "2026-06-14T00:00:00Z",
|
||
"codex_lineage": {
|
||
"segments": [
|
||
{
|
||
"id": "seg_0",
|
||
"start": 0,
|
||
"end": 5,
|
||
"start_byte": 0,
|
||
"end_byte": 54
|
||
}
|
||
]
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
## API Reference
|
||
|
||
### Symbolic Activation
|
||
|
||
```bash
|
||
POST /api/symbolic/activate
|
||
Content-Type: application/json
|
||
|
||
{
|
||
"intent": "I need creative image generation",
|
||
"request_type": "image",
|
||
"metrics": {
|
||
"power": 75,
|
||
"resonance": 70
|
||
}
|
||
}
|
||
|
||
Response:
|
||
{
|
||
"status": "success",
|
||
"glyph_id": "G002",
|
||
"specialized_type": "star_bloom_creativity",
|
||
"model": "forge",
|
||
"priority": 8.5,
|
||
"resonance_score": 87.3,
|
||
"power_boost": 14.50,
|
||
"superpower_count": 18,
|
||
"routing": {
|
||
"constraints": ["max_vram: 6.5GB"],
|
||
"enhancements": ["bloomflare_engine"],
|
||
"vram_budget": 6.5
|
||
}
|
||
}
|
||
```
|
||
|
||
### System Status
|
||
|
||
```bash
|
||
GET /api/status
|
||
|
||
Response:
|
||
{
|
||
"status": "operational",
|
||
"vram": {
|
||
"used_gb": 6.1,
|
||
"total_gb": 8.0,
|
||
"percent": 76.25
|
||
},
|
||
"vram_status": "VRAM safe",
|
||
"models_running": {
|
||
"llama": "available",
|
||
"forge": "available",
|
||
"janus": "pending",
|
||
"google_ai": "unconfigured"
|
||
},
|
||
"vram_mode": "8GB",
|
||
"compression": {
|
||
"enabled": true,
|
||
"format": "GSZ3",
|
||
"glyphmart": "ready"
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
## Performance Metrics
|
||
|
||
| Operation | Time | Notes |
|
||
|-----------|------|-------|
|
||
| Load 600 glyphs | <5ms | From JSON |
|
||
| Load 152 superpowers | <2ms | From JSON |
|
||
| Compress 1KB source | <1ms | GSZ3 + zlib |
|
||
| Decompress payload | <0.5ms | GSZ3 |
|
||
| Execute through LAIN | ~15ms | 8 lanes |
|
||
| Multi-glyph resonance | <50ms | 3 glyphs |
|
||
| Glyph activation | <10ms | Full pipeline |
|
||
|
||
---
|
||
|
||
## Testing
|
||
|
||
### Unit Tests
|
||
|
||
```bash
|
||
python3 tests/test_supercharged_registry.py
|
||
python3 tests/test_lain_glyph_bridge.py
|
||
python3 tests/test_cognitive_kernel.py
|
||
python3 tests/test_events.py
|
||
python3 tests/test_control_flow.py
|
||
```
|
||
|
||
### Integration Tests
|
||
|
||
```bash
|
||
python3 integration_tests/test_compile.py
|
||
python3 integration_tests/test_run.py
|
||
python3 integration_tests/test_inspect.py
|
||
python3 integration_tests/test_summary.py
|
||
python3 integration_tests/test_errors.py
|
||
python3 integration_tests/test_determinism.py
|
||
```
|
||
|
||
### Benchmark Tests
|
||
|
||
```bash
|
||
python3 benchmark/benchmark_superpowers.py
|
||
python3 benchmark/run_all_benchmarks.py
|
||
```
|
||
|
||
---
|
||
|
||
## Configuration
|
||
|
||
### Environment Variables
|
||
|
||
```bash
|
||
# VRAM mode
|
||
export VRAM_MODE="8GB" # 8GB, 24GB, 48GB
|
||
|
||
# External endpoints
|
||
export FEDMART_ENDPOINT="http://localhost:8000/fedmart/ingest/xic"
|
||
export TABBY_API="http://192.168.2.12:11436"
|
||
export GOOGLE_API_KEY="your_key_here"
|
||
```
|
||
|
||
### VRAM Configuration
|
||
|
||
```python
|
||
VRAM_WARNING = 6.5 # GB
|
||
VRAM_CRITICAL = 7.8 # GB
|
||
TOTAL_VRAM = 8.0 # GB (GTX 1080)
|
||
```
|
||
|
||
---
|
||
|
||
## Troubleshooting
|
||
|
||
### VRAM Critical
|
||
|
||
**Symptom**: OOM during pipeline load
|
||
|
||
**Solution**:
|
||
- Use `device_map="balanced"` mode
|
||
- Reduce batch size
|
||
- Close other models (Forge + Janus conflict)
|
||
|
||
### Compression Checksum Mismatch
|
||
|
||
**Symptom**: `GSZ3DecompressionError: Checksum mismatch`
|
||
|
||
**Solution**:
|
||
- Verify file integrity
|
||
- Re-compress source
|
||
- Check for file corruption
|
||
|
||
### Glyph Not Found
|
||
|
||
**Symptom**: `Glyph G001 not found`
|
||
|
||
**Solution**:
|
||
- Verify `supercharged_glyphs.json` exists
|
||
- Check file path in `super_registry.py`
|
||
- Run `glyph_explorer.py test`
|
||
|
||
---
|
||
|
||
## Future Enhancements
|
||
|
||
- [ ] Llama Chat integration (Pinokio)
|
||
- [ ] Janus Video generation (Pinokio)
|
||
- [ ] Google AI Vision (Gemini/Vertex)
|
||
- [ ] Database persistence for telemetry
|
||
- [ ] Authentication on endpoints
|
||
- [ ] Prometheus/Grafana metrics export
|
||
- [ ] PDF report generation
|
||
- [ ] Multi-run comparison view
|
||
|
||
---
|
||
|
||
**Version**: 2.0.0
|
||
**Build Date**: June 14, 2026
|
||
**Status**: ✅ Production Ready
|