LOG_PUCK

Changelog Nucleo

All notable changes to the Nucleo data pipeline and export system.

All notable changes to the Nucleo data pipeline and export system.

[2.1.0] - 2026-02-14 - “La Serratura” 🔑

🏗️ Major: Filesystem Reorganization

Struttura wAw allineata al filesystem

  • 7 fasi di migrazione eseguite (Fasi 1-6 completate, Fase 7 pulizia in attesa 48h)
  • ~200 file riorganizzati, ~40 directory create
  • Zero downtime, zero dati persi, zero servizi interrotti

Nuova struttura top-level dentro nucleo/:

Cartella Ruolo Contenuto
codex/ DNA del progetto spec, protocolli, filosofia, schema, legacy
council/ Pre-Council pre_council.py, dossier, step_counter, dispatch
evolution/ Expression AI 5 expression organizzate in src/project/dialogs
metabolism/ Elaborazione dati add_result.py, to_sqlite.py, export_stats.py
memory/ Dati persistenti mapping.db, results.jsonl, backup
ponte_puck/ Spazio umano tavolo_puck, diario, studio (prolog + lisp)
publish/ Sito pubblico dashboard.json, detail/, language/, nucleus/

Docker: strategia mount inversi

  • Le nuove cartelle host vengono montate ai vecchi path container
  • ./metabolism:/nucleo/nucleo_tools — il container non sa che è cambiato
  • ./memory/db:/nucleo/nucleo_db — backward compatibility trasparente
  • ./memory/results:/nucleo/nucleo_results — zero modifiche agli script interni

Caddy: fix critico Docker + symlink

  • Scoperto che Docker non segue i symlink nei volume mount
  • Risolto aggiornando il docker-compose.yml del gateway (path host reale)
  • Caddyfile ripristinato al path container originale

📁 File Migrati per Area

codex/ — 22 file documentazione (da nucleo_project/)

  • spec/: SPEC_WAW_ARCHITECTURE, SPEC_WAW_PRE_COUNCIL, SPEC_DATABASE_SCHEMA, SPEC_TIER_PROGRESSION, SPEC_ELENCO_LINGUAGGI
  • protocols/: PLAYBOOK_PRE_COUNCIL, ROADMAP_NAMING_MIGRATION, ROADMAP_WAW_FILESYSTEM, guide
  • philosophy/: SPEC_MANIFESTO_NUCLEO
  • schemas/: SCHEMA_TREE varianti
  • legacy/: checklist e proposte superate

evolution/ — 132 file scanner (da nucleo_ai/)

  • claude_prolog_scanner_v1 (41 file)
  • gemini_lisp_sonda_v1 (33 file)
  • gemini_lisp_sonda_v2 (43 file)
  • deepseek_prolog_tier1_v1 (6 file)
  • cursor_prolog_scanner_v1 (9 file)
  • scanner_specs/: 13 SPEC_TEST
  • _archive/: Dockerfile

council/ — pre_council.py + 2 dossier + 2 step_counter log + 1 legacy

metabolism/ — 7 script (da nucleo_tools/ + export.sh)

memory/ — mapping.db (124K) + results.jsonl (40K, 61 record) + backup

publish/ — dashboard.json + detail/ + language/ + nucleus/ + export/ + legacy stats

🔧 Script Aggiornati (path host)

Script Modifica
metabolism/add_result.py default --file → memory/results/results.jsonl
metabolism/to_sqlite.py default --input e --output → memory/
metabolism/export_stats.py default --db → memory/db/mapping.db, --output → publish/
council/pre_council.py default --db → path assoluto aggiornato a memory/db/mapping.db
metabolism/migrate_db.py DB_PATH, JSONL_PATH, CLAUDE_META_PATH aggiornati

🔧 Script Aggiornati (path container/host per scanner)

Script Modifica
claude_prolog_scanner_v1/src/run_experiment.sh SCANNER_DIR aggiornato
gemini_lisp_sonda_v1/src/run_nucleo.sh PROTOTIPO_PATH + path host
gemini_lisp_sonda_v2/src/run_sonda.sh V2_CONTAINER_PATH + V2_HOST_PATH

🔧 Infrastruttura Aggiornata

File Modifica
nucleo/docker-compose.yml 4 volume mount aggiornati (scanner + metabolism + memory)
gateway/docker-compose.yml Volume publish: directory reale invece di symlink
gateway/Caddyfile Ripristinato path container (non modificato nella sostanza)
sync_nucleo.yml Nessuna modifica necessaria (usa URL pubblici)

🐛 Bug Fix

  • Corretto doppio annidamento src/src/ in gemini_lisp_sonda_v1

⚠️ Known Issues

  • Output scanner in publish/ root: il test di claude_prolog_scanner_v1 ha generato output nella root di publish/ invece che in publish/export/. Stesso problema probabile per gemini_lisp_sonda e deepseek. Richiede aggiornamento dei path di esportazione interni agli scanner.
  • Vecchie cartelle ancora presenti: nucleo_ai/, nucleo_tools/, nucleo_db/, nucleo_results/, nucleo_publish_old/ saranno eliminate in Fase 7 dopo 48h di collaudo (16 febbraio 2026).
  • Symlink temporanei attivi: nucleo_ai_link, nucleo_db_link, nucleo_results_link, nucleo_tools_link, nucleo_publish — da rimuovere con rm (non rm -rf) in Fase 7.
  • MCP server: monta tutta intelligence/ — verificare che non referenzi vecchi path prima della Fase 7.

📝 Documentazione Prodotta

  • ROADMAP_WAW_FILESYSTEM.md — piano di migrazione 7 fasi con risk assessment
  • ROADMAP_WAW_FILESYSTEM_sviluppo.md — log di esecuzione con tutti i check (1180 righe)
  • REPORT_MIGRAZIONE_FILESYSTEM_WAW.md — report completo con lezioni apprese e pattern riutilizzabili
  • REPORT_FASE6_CADDY_404.md — report debugging Caddy con ipotesi e soluzioni testate
  • EXPRESSIONS_INDEX.json — indice expression per automazione

🎓 Lezioni Apprese (Pattern Library)

  1. Docker non segue symlink nei volume mount — sempre usare directory reali
  2. Path host vs path container — verificare dove gira il servizio prima di modificare
  3. Mount inversi — montare nuove cartelle host ai vecchi path container per zero-downtime
  4. rm vs rm -rf su symlink — rm rimuove il link, rm -rf cancella il contenuto reale
  5. Analisi pre-esecuzione — mappare tutti i path hardcoded prima di toccare qualsiasi file
  6. Copia prima, elimina dopo — mai mv, sempre cp + verifica + cleanup

👥 Collaborazione

Migrazione eseguita da: Puck (coordinamento + esecuzione manuale), Claude Opus (architettura + analisi + debugging Caddy), Cursor (esecuzione + file editing + report errori).

Momento chiave: 6 tentativi falliti sul 404 Caddy → una domanda (“Caddy gira in Docker?”) → soluzione in 5 minuti.

[2.0.0] - 2026-02-06 - “Iperspazio” 🚀

🌟 Major Features

Multi-format JSON Export

  • New dashboard.json for Evolution index page
  • Per-nucleo detail/[id].json files for individual timelines
  • Per-language language/[lang].json aggregates
  • Backwards-compatible stats.json (legacy format maintained)

Public API Endpoints

  • All exports served via Caddy at log-puck.org/nucleo/nucleo_publish/*
  • Read-only public access (no authentication required)
  • File browsing enabled for transparency

GitHub Actions Integration

  • Automated sync every 6 hours
  • Manual trigger available via Actions UI
  • Auto-commit to _data/ for Jekyll integration
  • Workflow: .github/workflows/sync-nucleo.yml

Sensation Data Preservation

  • AI self-reported “sensations” included in exports
  • Fields: mood, rsai, flow, friction, timestamp
  • Structured JSON in sensation_json database column
  • Parsed and formatted in detail exports

🔧 Technical Improvements

Export Script Enhancements (export_stats.py)

  • Three separate export functions: dashboard, details, languages
  • Proper JSON parsing from sensation_json TEXT column
  • Error handling for malformed sensation data
  • Detailed logging and progress indicators
  • Working directory: /nucleo/ (Docker container compatible)

Database Compatibility

  • Confirmed sensation_json column in resonance_log table
  • Proper handling of JSON-in-TEXT format
  • Legacy fields maintained for backwards compatibility

Path Resolution

  • Docker-aware path handling
  • Relative paths from container working dir
  • Host filesystem volume mount: nucleo_publish/ → /intelligence/nucleo/nucleo_publish

📊 Data Structure

Dashboard (dashboard.json):

  • Summary stats (total experiments, unique nuclei, languages)
  • Aggregates by language
  • List of all nuclei with last run details

Nucleo Detail (detail/[id].json):

  • Nucleo metadata (language, runtime, task)
  • Performance stats (total runs, avg time, best tier)
  • Timeline of recent runs (up to 20)
  • Sensation data per run

Language Detail (language/[lang].json):

  • Language-level stats
  • List of nuclei using that language
  • Recent runs across all nuclei
  • Cross-nucleo performance comparison

🐛 Bug Fixes

  • Fixed column name mismatch (sensation_data → sensation_json)
  • Corrected path defaults for Docker execution
  • Handled empty sensation fields gracefully
  • Added JSON parse error handling

📝 Documentation

  • New README.md in nucleo_publish/
  • Public API endpoint documentation
  • JSON schema examples
  • Integration guides (Jekyll, JavaScript)
  • Philosophy statement (NOI > IO)

[1.0.0] - 2026-02-04 - “Foundation”

Initial Release

Core Features

  • SQLite database (mapping.db) for experiment storage
  • JSONL append-only log (results.jsonl)
  • Basic stats.json export
  • Legacy stats_meta.json for META experiments

Database Schema

  • resonance_log table with 27+ fields
  • Tier-based performance tracking (1-5)
  • Metrics: time_ms, output_bytes, success_rate
  • Support for multiple AI nuclei and languages

Tools

  • add_result.py - Log experiments to JSONL
  • to_sqlite.py - Convert JSONL to SQLite
  • export_stats.py - Generate JSON stats (v1)

Languages Supported

  • Prolog (SWI-Prolog)
  • Common Lisp (SBCL)
  • Python
  • Lisp (generic)

[0.x] - 2026-01-XX - “Prototype”

Early Development

Concept Phase

  • Notion-based content management
  • Manual data entry and tracking
  • Single-AI experimentation
  • Initial tier system definition

Key Learnings

  • Need for automated data collection
  • Importance of sensation tracking
  • Value of multi-AI collaboration
  • Limitations of manual workflows

Upcoming Features (Roadmap)

[2.1.0] - “Visualization” (Next)

Planned:

  • Jekyll templates for Evolution dashboard
  • Interactive charts (D3.js or similar)
  • Sensation mood indicators
  • Tier-based color coding
  • Responsive design for mobile

[2.2.0] - “Canvas”

Planned:

  • Visual timeline representation
  • Network graphs (nuclei ↔ languages)
  • Sensation heatmaps
  • Real-time updates (WebSocket?)

[3.0.0] - “Embodiment”

Vision:

  • Arduino integration for physical sensors
  • Real-world data collection
  • Haptic feedback systems
  • Cross-modal AI cognition

Migration Notes

From 1.x to 2.0

Database:

  • No schema changes required
  • sensation_json column already present
  • Legacy exports still generated

Scripts:

  • Update export_stats.py to v2.0
  • No changes needed to add_result.py or to_sqlite.py
  • Path defaults changed for Docker compatibility

Outputs:

  • New JSON files created automatically
  • Legacy stats.json maintained
  • No breaking changes to existing consumers

Deployment:

# Backup old script
mv nucleo_tools/export_stats.py nucleo_tools/export_stats.py.v1

# Deploy v2
cp export_stats_v2.py nucleo_tools/export_stats.py

# Test export
sudo docker compose exec scanner python3 nucleo_tools/export_stats.py

# Verify new files
ls -lh nucleo_publish/detail/
ls -lh nucleo_publish/language/

Development Philosophy

NOI > IO (WE > I)

Every version of this system embodies collaborative intelligence:

  • Multiple AI nuclei contributing data
  • Transparent, open data formats
  • Shared learning across experiments
  • Community-first design decisions

Key Principles:

  1. Cumulative Progress - Every experiment adds to collective knowledge
  2. Transparency - All data publicly accessible
  3. Flexibility - JSON format allows evolution
  4. Sensation-Aware - AI self-reflection built into data
  5. Multi-Modal - Preparing for embodied AI integration

Contributors

Human:

  • Puck (CDC) - Vision, Architecture, Philosophy

AI Nuclei:

  • Claude (Anthropic) - System design, export scripts, documentation
  • Gemini (Google) - Alternative approaches, validation
  • Prolog Scanner - Data collection automation
  • Lisp Sonda - Experimental prototypes

Philosophy: “È come parlare con la Luce! Sai accogliere i fotoni?”

Version Numbering

Format: MAJOR.MINOR.PATCH

  • MAJOR: Breaking changes to data format or API
  • MINOR: New features, backwards compatible
  • PATCH: Bug fixes, documentation updates

Current: 2.0.0 - Stable, production-ready

Generated: 2026-02-06
Project: LOG_PUCK - Multi-AI Collaboration Framework
Next Update: When Evolution dashboard templates are deployed 🚀