Skip to content

AI-INTEGRATION-GUIDE.md

Latest commit

 

History

History
1109 lines (826 loc) · 27.8 KB

AI-INTEGRATION-GUIDE.md

File metadata and controls

1109 lines (826 loc) · 27.8 KB
title modules/KENL AI Integration Guide
version 1.0.0
date 2025-11-10
classification OWI-DOC
atom ATOM-DOC-20251110-018

modules/KENL AI Integration Guide

How AI assistants (Claude, Qwen, Perplexity) enhance each modules/KENL module

AI Assistance Levels

graph LR
    A[No AI Needed] -->|Basic commands| B[Little Help]
    B -->|Explanations| C[Optional Help]
    C -->|Recommendations| D[Recommended]
    D -->|Core functionality| E[Maximum Help]
    E -->|Required| F[AI Required]

    style A fill:#f8f9fa,stroke:#495057
    style B fill:#e7f5ff,stroke:#228be6
    style C fill:#d0ebff,stroke:#228be6
    style D fill:#a5d8ff,stroke:#228be6
    style E fill:#74c0fc,stroke:#228be6
    style F fill:#4dabf7,stroke:#228be6
Loading

modules/KENL AI Integration Matrix

modules/KENL AI Level Why Best AI Model
KENL0 System 🟦 Little Help System commands are straightforward, AI explains complex operations Qwen 2.5 (local) - Fast, offline
KENL1 Framework 🟨 Optional Help ATOM trail analysis, recovery recommendations Claude - Best reasoning
KENL2 Gaming 🟩 Maximum Help ProtonDB research, compatibility, optimization Perplexity - Web research
KENL3 Dev 🟩 Maximum Help Code generation, debugging, container setup Claude Code - Programming
KENL4 Monitoring 🟨 Optional Help Metric analysis, anomaly detection Qwen 2.5 - Fast local inference
KENL5 Facades 🟦 Little Help Config files, visual theming Qwen 2.5 - Simple tasks
KENL6 Social 🟨 Optional Help Generate Play Card descriptions Claude - Creative writing
KENL7 Learning 🟩 Maximum Help Tutoring, examples, concept explanations Claude + Perplexity - Teaching
KENL8 Security 🟧 Recommended Best practices, but verify all suggestions Claude - Security knowledge
KENL9 Library 🟨 Optional Help Storage optimization, migration planning Qwen 2.5 - Fast local
KENL10 Backup 🟦 Little Help Mostly automated, AI can schedule Qwen 2.5 - Fast local
KENL11 Media 🟩 Maximum Help Troubleshooting, quality profiles, indexer config Perplexity + Claude - Research + reasoning

Legend:

  • 🟦 Little Help (10-20% AI usage) - AI explains, you execute
  • 🟨 Optional Help (30-50% AI usage) - AI recommends, you decide
  • 🟧 Recommended (60-70% AI usage) - AI guides, you verify
  • 🟩 Maximum Help (80-90% AI usage) - AI automates, you approve
  • 🟥 AI Required (100% AI) - Cannot function without AI

Detailed AI Integration by modules/KENL

modules/KENL0: System Operations 🟦 (Little Help)

AI Usage: 10-20%

What AI helps with:

  • Explaining complex rpm-ostree concepts
  • Translating ujust recipes into plain English
  • Suggesting rollback strategies when updates fail

Example workflow:

# You run a command
⚙️ modules/KENL0 $ rpm-ostree status

# Output is confusing
State: idle
Deployments:
● ostree-image-signed:docker://quay.io/bazzite/...

# Ask AI to explain
You: "What does this deployment output mean?"

AI (Qwen): "You're on an OCI-based deployment. The '●' means
it's currently active. This is Bazzite pulling from Quay.io
container registry. The image is signed for security."

Best AI model: Qwen 2.5 14B (local)

  • Why: Fast responses for quick explanations
  • Where: Runs locally via Ollama on /mnt/claude-ai/ (KENL9 partition)
  • Privacy: 100% offline, no data sent to cloud

Setup:

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull Qwen model
ollama pull qwen2.5:14b

# Move to external drive (KENL9)
mv ~/.ollama/models /mnt/claude-ai/models/ollama
ln -s /mnt/claude-ai/models/ollama ~/.ollama/models

# Use in modules/KENL0
echo "alias ai='ollama run qwen2.5:14b'" >> ~/.bashrc

AI prompts for modules/KENL0:

"Explain this rpm-ostree error: [paste error]"
"What's the difference between rpm-ostree upgrade and rebase?"
"How do I rollback if update breaks my system?"
"What does 'layered packages' mean in Bazzite?"

modules/KENL1: Framework Core 🟨 (Optional Help)

AI Usage: 30-50%

What AI helps with:

  • Analyzing ATOM trail for recovery insights
  • Suggesting next actions based on intent history
  • Correlating crashes with recent changes

Example workflow:

# System crashed, you have ATOM trail
⚛️ modules/KENL1 $ atom-analytics --recovery

# AI analyzes trail
AI (Claude): "Based on your ATOM trail:

ATOM-CFG-20251109-045: Updated Proton GE to 9-20
ATOM-GAMING-20251109-046: Launched Halo Infinite
ATOM-CRASH-20251109-047: System crash (GPU driver)

Timeline suggests Proton GE 9-20 incompatible with your
NVIDIA 570.86.10 driver. Recommend:
1. Rollback Proton to 9-18
2. Or update NVIDIA driver to 575.x
3. Test in modules/KENL2 before marking as resolved"

Best AI model: Claude 3.5 Sonnet (via API)

  • Why: Best reasoning and pattern recognition
  • Privacy: Sends ATOM trail summaries (no personal data)
  • Cost: ~$0.01 per recovery analysis

Setup:

# Configure Claude API
export ANTHROPIC_API_KEY="your-key-here"

# Add to modules/KENL1
cat >> ~/.kenl/KENL1-framework/config <<EOF
[ai]
provider = anthropic
model = claude-3-5-sonnet-20241022
max_tokens = 4000
EOF

AI prompts for modules/KENL1:

"Analyze this ATOM trail and suggest recovery: [paste trail]"
"What was I doing before the crash?"
"Find patterns in my last 100 ATOM entries"
"Suggest next action based on my recent intent"

modules/KENL2: Gaming 🟩 (Maximum Help)

AI Usage: 80-90%

What AI helps with:

  • ProtonDB research (Perplexity searches compatibility)
  • Automatic Play Card generation (Claude analyzes configs)
  • Optimization recommendations (AI suggests best Proton version)
  • Troubleshooting (AI correlates errors with known issues)

Example workflow (fully AI-assisted):

# You want to play a new game
🎮 modules/KENL2 $ setup-game "Baldur's Gate 3"

# AI takes over (Perplexity + Claude)
AI Research (Perplexity):
  🔍 Searching ProtonDB for "Baldur's Gate 3"...
  ✅ Platinum rating (9.3/10, 3,421 reports)

  Best configuration:
    - Proton: GE-Proton 9-18
    - Launch options: PROTON_USE_WINED3D=1 %command%
    - DXVK: 2.3
    - Issues: None for NVIDIA RTX 3080

AI Configuration (Claude):
  📝 Generating Play Card...

  hardware:
    gpu: NVIDIA RTX 3080 (detected)
    cpu: AMD Ryzen 7 5800X3D (detected)
    ram: 32GB (detected)

  configuration:
    proton: GE-Proton 9-18
    launch_options: "PROTON_USE_WINED3D=1 %command%"
    dxvk_version: 2.3

  expected_performance:
    resolution: 1440p
    settings: Ultra
    fps_estimate: 90-110 FPS

  ✅ Apply this configuration? [Y/n]

You just type: y → Everything configured automatically!

Best AI models: Perplexity (research) + Claude (configuration)

Setup:

# Perplexity API for ProtonDB research
export PERPLEXITY_API_KEY="your-key-here"

# Claude for Play Card generation
export ANTHROPIC_API_KEY="your-key-here"

# Configure modules/KENL2
cat >> ~/.kenl/KENL2-gaming/config <<EOF
[ai]
research_provider = perplexity
research_model = llama-3.1-sonar-large-128k-online
config_provider = anthropic
config_model = claude-3-5-sonnet-20241022
auto_research = true
auto_configure = true
EOF

AI prompts for modules/KENL2:

"Research Elden Ring Linux compatibility"
"Generate Play Card from my Steam appmanifest"
"Why is Halo Infinite stuttering? Check ProtonDB"
"Compare Proton GE 9-18 vs 9-20 for Cyberpunk"
"Optimize my Play Card for 4K 60fps"

AI-generated content:

  • Play Card templates
  • Launch option recommendations
  • Performance tuning guides
  • Compatibility reports

modules/KENL3: Development 🟩 (Maximum Help)

AI Usage: 80-90%

What AI helps with:

  • Code generation (Claude Code writes complete functions)
  • Debugging (AI analyzes stack traces)
  • Container setup (AI writes Dockerfiles/distrobox configs)
  • Git workflows (AI suggests commit messages, resolves conflicts)

Example workflow:

# You need a Python ML environment
💻 modules/KENL3 $ create-devbox python-ml

# AI generates complete setup (Claude Code)
AI: "Creating Python ML development environment...

Dockerfile generated:
FROM ubuntu:24.04

RUN apt-get update && apt-get install -y \
    python3.12 python3-pip python3-venv \
    build-essential git curl

RUN pip install --no-cache-dir \
    torch torchvision torchaudio \
    numpy pandas scikit-learn \
    jupyter matplotlib seaborn

WORKDIR /workspace

Distrobox config:
  name: python-ml
  image: ubuntu:24.04
  volumes:
    - ~/code:/workspace
    - /mnt/claude-ai:/claude-ai:ro

Container created! Enter with: dbe python-ml"

Best AI model: Claude Code (this conversation!)

  • Why: Best for programming tasks
  • Integration: Already running in your modules/KENL3 distrobox

AI prompts for modules/KENL3:

"Write a Python script to analyze ATOM trail JSON"
"Debug this containerization error: [paste error]"
"Generate Dockerfile for Rust development"
"Explain this git merge conflict and suggest resolution"
"Write unit tests for this function: [paste code]"

modules/KENL4: Monitoring 🟨 (Optional Help)

AI Usage: 30-50%

What AI helps with:

  • Anomaly detection (AI spots unusual FPS drops)
  • Performance recommendations (AI suggests optimizations)
  • Metric correlation (AI finds why temps spiked)

Example workflow:

# You notice FPS drops
📊 modules/KENL4 $ kenl-monitor analyze --last 7d

# AI analyzes metrics (Qwen local)
AI: "Performance anomaly detected:

Pattern identified:
  - FPS drops from 90 → 45 every ~30 minutes
  - Correlates with GPU temp spike (65°C → 85°C)
  - Happens during Cyberpunk sessions only

Root cause: Thermal throttling
  Your RTX 3080 hits thermal limit, GPU clock
  drops from 1900MHz → 1200MHz

Recommendation:
  1. Improve case airflow
  2. Adjust fan curve (more aggressive)
  3. Limit FPS to 75 (reduce heat generation)
  4. Consider undervolting GPU"

Best AI model: Qwen 2.5 14B (local)

  • Why: Fast, can run continuously monitoring metrics
  • Privacy: All analysis happens locally

modules/KENL7: Learning 🟩 (Maximum Help)

AI Usage: 80-90%

What AI helps with:

  • Tutoring (AI teaches Git, rpm-ostree, GPG step-by-step)
  • Generating examples (AI creates runnable demos)
  • Explaining concepts (AI breaks down complex topics)
  • Interactive Q&A (AI answers follow-up questions)

Example workflow:

# You want to learn Git rebase
🎓 modules/KENL7 $ learn git-rebase

# AI tutors you (Claude + Perplexity)
AI (Claude): "Let's learn Git rebase interactively!

Git rebase rewrites commit history to make it linear.

Example scenario:
  main:    A---B---C
            \
  feature:  D---E

After rebase:
  main:    A---B---C
                    \
  feature:           D'---E'

Let's practice with a real example...

[Generates sandbox git repo with commits]
[Walks through rebase step-by-step]
[Explains each step with visuals]
[Asks comprehension questions]

Quiz: What's the difference between rebase and merge?
You: [answer]
AI: Correct! Rebase rewrites history, merge preserves it.

Next: Try rebasing your feature branch on updated main!"

Best AI models: Claude (teaching) + Perplexity (research)

AI-generated content:

  • Interactive tutorials
  • Runnable code examples
  • Visual diagrams
  • Practice exercises
  • Quizzes

modules/KENL8: Security 🟧 (Recommended - Verify All Suggestions!)

AI Usage: 60-70%

What AI helps with:

  • Security best practices
  • GPG key management guidance
  • Vault integration setup
  • ⚠️ WARNING: AI can make dangerous suggestions!

Example workflow:

# You want to setup GPG
🔐 modules/KENL8 $ setup-gpg

# AI guides you (Claude) - YOU VERIFY EACH STEP
AI: "Setting up GPG for commit signing...

Step 1: Generate key
  Command: gpg --full-generate-key

  Recommended settings:
    - Type: (4) RSA (sign only)
    - Size: 4096 bits
    - Expiration: 2 years

  ⚠️ VERIFY: Does this match security best practices?

Step 2: Export public key
  Command: gpg --armor --export your-email@example.com

  ⚠️ VERIFY: Never export private key!

Step 3: Add to GitHub
  [Instructions...]

⚠️ AI-GENERATED SECURITY ADVICE ⚠️
Always verify with official documentation!"

Best AI model: Claude 3.5 Sonnet

  • Why: Good security knowledge
  • ⚠️ Always verify: Check official docs, don't blindly trust AI!

Proton Setup (VPN + Mail + Drive) - AI Assisted:

# Setup Proton VPN
🔐 modules/KENL8 $ setup-proton-vpn

AI: "Proton VPN setup for privacy-focused torrenting...

Step 1: Install Proton VPN client
  Method 1 (GUI): flatpak install protonvpn
  Method 2 (CLI): Download from protonvpn.com/download

Step 2: Login with Proton account
  Username: your-email@proton.me
  Password: [use Proton password, not account password]

Step 3: Configure for torrenting
  - Enable Kill Switch (prevents IP leaks)
  - Enable DNS leak protection
  - Select P2P-optimized servers

  Recommended servers for P2P:
    🇳🇱 Netherlands (nl-free-01.protonvpn.net)
    🇨🇭 Switzerland (ch-free-01.protonvpn.net)
    🇸🇪 Sweden (se-free-01.protonvpn.net)

Step 4: Test connection
  curl ifconfig.me
  Expected: Proton VPN IP (not your real IP)

Step 5: Integrate with modules/KENL11 (qBittorrent)
  See: ~/kenl/KENL11-media/vpn/setup-proton.sh

✅ Proton VPN configured for secure torrenting!"

Proton Mail + Drive setup:

# Setup Proton Mail Bridge (for local clients)
🔐 modules/KENL8 $ setup-proton-mail

AI: "Proton Mail Bridge allows Thunderbird/Mutt to access
encrypted Proton Mail locally.

Step 1: Install Bridge
  flatpak install protonmail-bridge

Step 2: Login and get IMAP/SMTP credentials
  Bridge generates local credentials for email clients

Step 3: Configure Thunderbird
  IMAP: 127.0.0.1:1143 (local Bridge)
  SMTP: 127.0.0.1:1025 (local Bridge)
  Username: your-email@proton.me
  Password: [Bridge-generated password]

✅ Encrypted email locally accessible!"

# Setup Proton Drive
🔐 modules/KENL8 $ setup-proton-drive

AI: "Proton Drive for encrypted cloud storage...

Step 1: Install Proton Drive client
  Download: proton.me/drive/download
  Or: Use web interface (drive.proton.me)

Step 2: Sync folders
  Recommend syncing:
    - modules/KENL10 snapshots (encrypted backups)
    - Play Cards (share with friends)
    - Important documents

Step 3: Integrate with modules/KENL10 backup
  See: ~/kenl/KENL10-backup/cloud-sync/proton-drive.sh

✅ Encrypted cloud storage configured!"

⚠️ Security verification checklist (AI-generated):

Before executing AI suggestions:
- [ ] Check official Proton documentation
- [ ] Verify commands don't expose credentials
- [ ] Test VPN kill switch works
- [ ] Confirm DNS leak protection active
- [ ] Validate GPG key settings match best practices
- [ ] Ensure private keys never leave your machine

modules/KENL11: Media 🟩 (Maximum Help)

AI Usage: 80-90%

What AI helps with:

  • Troubleshooting indexers (AI fixes Prowlarr errors)
  • Quality profile recommendations (AI suggests bitrates)
  • Automatic subtitle matching (AI finds best subs)
  • Media organization (AI suggests folder structures)

Example workflow:

# Prowlarr indexer failing
📺 modules/KENL11 $ troubleshoot-indexer "1337x"

# AI debugs (Perplexity + Claude)
AI (Perplexity): "Researching 1337x recent issues...

Found: 1337x changed domain (1337x.to → 1337xx.to)
Community reports: Cloudflare protection increased

AI (Claude): "Fixing indexer configuration...

Changes needed:
1. Update URL: 1337x.to → 1337xx.to
2. Enable FlareSolverr (Cloudflare bypass)
3. Increase timeout: 30s → 60s

Applying fixes...
  ✅ Prowlarr indexer updated
  ✅ FlareSolverr configured
  ✅ Testing... Success!

1337x is working again. Syncing to Radarr/Sonarr..."

Best AI models: Perplexity (research) + Claude (fix)

AI prompts for modules/KENL11:

"Why is my indexer failing? Check recent changes"
"Recommend quality profile for 1080p streaming"
"Troubleshoot: Radarr not finding any releases"
"Setup Proton VPN for qBittorrent container"
"Optimize Jellyfin transcoding for my GPU"

AI Privacy Tiers

Tier 1: 100% Local (Max Privacy) 🟢

AI Models:

  • Qwen 2.5 14B (via Ollama)
  • Llama 3.1 8B (via Ollama)

Storage: /mnt/claude-ai/models/ollama/ (KENL9 partition)

KENLs using local AI:

  • modules/KENL0 (system explanations)
  • modules/KENL4 (metric analysis)
  • modules/KENL5 (config generation)
  • modules/KENL9 (storage optimization)
  • modules/KENL10 (backup scheduling)

Setup:

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull models
ollama pull qwen2.5:14b
ollama pull llama3.1:8b

# Move to external drive
mv ~/.ollama/models /mnt/claude-ai/models/ollama
ln -s /mnt/claude-ai/models/ollama ~/.ollama/models

# Use in any modules/KENL
ollama run qwen2.5:14b "Explain rpm-ostree"

Pros:

  • ✅ 100% offline, no data sent to cloud
  • ✅ Free (no API costs)
  • ✅ Fast (local inference)
  • ✅ Unlimited queries

Cons:

  • ❌ Requires powerful GPU (8GB+ VRAM recommended)
  • ❌ Slower than cloud models
  • ❌ Less capable than Claude/GPT-4

Tier 2: Cloud API (Moderate Privacy) 🟡

AI Models:

  • Claude 3.5 Sonnet (Anthropic API)
  • Perplexity Sonar (Perplexity API)

KENLs using cloud AI:

  • modules/KENL1 (ATOM trail analysis)
  • modules/KENL2 (ProtonDB research)
  • modules/KENL3 (code generation)
  • modules/KENL6 (Play Card descriptions)
  • modules/KENL7 (tutoring)
  • modules/KENL8 (security guidance - verify!)
  • modules/KENL11 (indexer troubleshooting)

Setup:

# Get API keys
# Claude: console.anthropic.com
# Perplexity: perplexity.ai/settings/api

# Configure globally
cat >> ~/.bashrc <<EOF
export ANTHROPIC_API_KEY="sk-ant-..."
export PERPLEXITY_API_KEY="pplx-..."
EOF

# Use in modules/KENLs
claude-api "Analyze this ATOM trail: [...]"
perplexity-api "Research Elden Ring Linux compatibility"

Pros:

  • ✅ Best AI capabilities
  • ✅ Fast responses
  • ✅ Up-to-date knowledge (Perplexity searches web)

Cons:

  • ❌ Costs money (~$0.01-0.10 per query)
  • ❌ Sends data to cloud (anonymized)
  • ❌ Requires internet

Privacy notes:

  • modules/KENL ATOM trail summaries are anonymized before sending
  • No personal identifiers sent
  • Can configure to redact sensitive data

Tier 3: Hybrid (Best of Both) 🟢🟡

Strategy:

  • Local AI for simple queries (Qwen)
  • Cloud AI for complex reasoning (Claude)

Auto-routing logic:

def choose_ai(query, kenl_context):
    # Simple queries → Local AI (free, fast)
    if is_simple_query(query):
        return ollama_qwen()

    # Complex reasoning → Cloud AI (best quality)
    if requires_research(query):
        return perplexity_api()

    if requires_coding(query):
        return claude_api()

    # Security queries → Local AI (privacy)
    if kenl_context == "KENL8":
        return ollama_qwen()

Best for:

  • Cost-conscious users
  • Privacy-focused users
  • Users wanting best of both worlds

Recommended AI Setup by Use Case

1. Privacy-First User

Goal: Never send data to cloud

Setup:

# Local AI only
ollama pull qwen2.5:14b
ollama pull llama3.1:70b  # Larger model for better quality

# Configure all modules/KENLs to use local
for kenl in ~/.kenl/KENL*/config; do
  echo "ai_provider = ollama" >> $kenl
  echo "ai_model = qwen2.5:14b" >> $kenl
done

Limitations:

  • Slower responses
  • Less capable for complex tasks
  • No web research (Perplexity equivalent)

2. Cost-Conscious User

Goal: Minimize AI API costs

Setup:

# Local AI for 80% of queries
ollama pull qwen2.5:14b

# Cloud AI only for complex tasks
export ANTHROPIC_API_KEY="..."  # Paid API

# Auto-route based on complexity
ai-router --local-first --cloud-fallback

Cost estimate:

  • Local: $0 (free)
  • Cloud: ~$5-10/month (occasional Claude queries)

3. Power User

Goal: Best AI assistance possible

Setup:

# All AI models
ollama pull qwen2.5:14b  # Local fallback
export ANTHROPIC_API_KEY="..."  # Claude
export PERPLEXITY_API_KEY="..."  # Perplexity
export OPENAI_API_KEY="..."  # GPT-4 (optional)

# Use best model for each task
ai-router --best-quality

Cost estimate:

  • ~$20-50/month (heavy AI usage across all modules/KENLs)

AI Safety Guidelines

✅ Safe AI Usage

KENL0 (System):

  • ✅ Ask AI to explain commands
  • ✅ Have AI generate documentation
  • ⚠️ Always read commands before executing!

KENL2 (Gaming):

  • ✅ AI researches ProtonDB
  • ✅ AI generates Play Cards
  • ✅ AI suggests optimizations
  • ⚠️ Test configs before sharing with friends

KENL3 (Dev):

  • ✅ AI writes code
  • ✅ AI debugs errors
  • ⚠️ Review generated code (security, efficiency)

KENL7 (Learning):

  • ✅ AI teaches concepts
  • ✅ AI generates examples
  • ✅ AI answers questions
  • ⚠️ Verify with official docs for critical topics

⚠️ Dangerous AI Usage

KENL0 (System):

  • ❌ Blindly running AI-generated system commands
  • ❌ Letting AI modify bootloader without review
  • ❌ AI-generated firmware update scripts

KENL8 (Security):

  • ❌ Trusting AI-generated GPG keys without verification
  • ❌ AI-suggested security configs without testing
  • ❌ Letting AI manage passwords/credentials

KENL11 (Media):

  • ❌ AI selecting torrents without virus scan
  • ❌ Blindly trusting AI indexer recommendations
  • ❌ AI managing VPN without kill switch verification

AI Configuration Files

Global AI Config

# ~/.config/kenl-ai/config.yaml

ai:
  # Default provider
  default_provider: hybrid  # ollama, anthropic, perplexity, hybrid

  # Providers
  providers:
    ollama:
      host: http://localhost:11434
      model: qwen2.5:14b
      timeout: 60

    anthropic:
      api_key: ${ANTHROPIC_API_KEY}
      model: claude-3-5-sonnet-20241022
      max_tokens: 8000

    perplexity:
      api_key: ${PERPLEXITY_API_KEY}
      model: llama-3.1-sonar-large-128k-online

  # Routing rules
  routing:
    simple_queries: ollama      # Fast, cheap
    complex_queries: anthropic  # Best reasoning
    research_queries: perplexity  # Web search
    security_queries: ollama    # Privacy

  # Privacy
  privacy:
    anonymize: true
    redact_patterns:
      - email
      - api_key
      - password
    local_only_kenls:
      - modules/KENL8  # Security - never send to cloud

Per-KENL AI Config

# ~/.kenl/KENL2-gaming/ai-config.yaml

kenl2_ai:
  # Gaming-specific AI settings
  protondb_research:
    provider: perplexity
    auto_research: true
    cache_results: true
    cache_ttl: 7d

  playcard_generation:
    provider: anthropic
    auto_generate: true
    include_hardware: true
    include_performance: true

  optimization:
    provider: ollama  # Fast local recommendations
    auto_optimize: false  # Ask before applying

Proton Services Integration

Proton VPN (Privacy-First Torrenting)

Integration with modules/KENL11-media:

# Setup Proton VPN for qBittorrent
📺 modules/KENL11 $ setup-proton-vpn

AI: "Configuring Proton VPN for torrenting...

Step 1: Install Proton VPN
  flatpak install flathub com.protonvpn.www

Step 2: Get Proton VPN credentials
  Login: protonvpn.com → Account → OpenVPN/IKEv2

  Credentials:
    Username: <username>+pmp  # +pmp for port forwarding
    Password: <openvpn-password>

Step 3: Configure qBittorrent container
  docker-compose up with Proton VPN sidecar:

  services:
    protonvpn:
      image: qmcgaw/gluetun
      environment:
        - VPN_SERVICE_PROVIDER=protonvpn
        - OPENVPN_USER=<username>+pmp
        - OPENVPN_PASSWORD=<password>
        - SERVER_COUNTRIES=Netherlands
        - FIREWALL_OUTBOUND_SUBNETS=192.168.1.0/24
      cap_add:
        - NET_ADMIN

    qbittorrent:
      network_mode: service:protonvpn  # Routes through VPN
      depends_on:
        - protonvpn

Step 4: Test kill switch
  ./vpn/test-killswitch.sh

  Expected: Traffic stops if VPN drops ✅

✅ Proton VPN configured for secure torrenting!"

docker-compose.proton.yml created: /home/user/kenl/KENL11-media/docker-compose/docker-compose.proton.yml

Proton Mail (Encrypted Email)

Integration with modules/KENL8-security:

# Setup Proton Mail Bridge
🔐 modules/KENL8 $ setup-proton-mail

AI: "Proton Mail Bridge for local email clients...

Use cases:
  - Encrypted email with Thunderbird/Mutt
  - Automated notifications from modules/KENL services
  - Sharing Play Cards via encrypted email

Step 1: Install Bridge
  flatpak install flathub ch.protonmail.protonmail-bridge

Step 2: Configure Bridge
  protonmail-bridge --cli
  > login
  > info

  IMAP: 127.0.0.1:1143
  SMTP: 127.0.0.1:1025
  Password: [Bridge-generated]

Step 3: Configure Thunderbird
  Account: your-email@proton.me
  IMAP: 127.0.0.1:1143 (SSL/TLS, Normal password)
  SMTP: 127.0.0.1:1025 (SSL/TLS, Normal password)

Step 4: Integrate with modules/KENL6 (social sharing)
  Share Play Cards via encrypted email:

  kenl-switch 6
  share-playcard halo.yaml friend@proton.me --method email

  → Email sent via Proton Mail (E2E encrypted)

✅ Encrypted email configured!"

Proton Drive (Encrypted Cloud Backup)

Integration with modules/KENL10-backup:

# Setup Proton Drive for cloud backups
💾 modules/KENL10 $ setup-proton-drive

AI: "Proton Drive for encrypted cloud backups...

Storage tiers:
  - Free: 5GB (sufficient for configs)
  - Plus: 500GB ($4/month)
  - Unlimited: 3TB ($10/month)

Step 1: Install Proton Drive client
  # Currently web-only, no Linux client yet
  # Workaround: Use rclone with Proton Drive

Step 2: Configure rclone
  rclone config
  > n (new remote)
  > name: proton-drive
  > type: webdav
  > url: https://drive.proton.me/dav
  > user: your-email@proton.me
  > pass: [app-specific password]

Step 3: Test connection
  rclone ls proton-drive:

Step 4: Integrate with modules/KENL10
  Automatic snapshots → Proton Drive:

  ~/kenl/KENL10-backup/cloud-sync/proton-drive.sh:
    - Snapshot created → Encrypt → Upload to Proton Drive
    - E2E encrypted in transit and at rest
    - Zero-knowledge (Proton can't read your backups)

✅ Encrypted cloud backup configured!"

Auto-sync script created: /home/user/kenl/KENL10-backup/cloud-sync/proton-drive.sh

Summary: AI Usage Recommendations

modules/KENL Recommended AI Privacy Level Cost/Month
KENL0 Qwen (local) 🟢 High $0
KENL1 Claude 🟡 Medium $1-2
KENL2 Perplexity + Claude 🟡 Medium $5-10
KENL3 Claude Code 🟡 Medium Included
KENL4 Qwen (local) 🟢 High $0
KENL5 Qwen (local) 🟢 High $0
KENL6 Claude 🟡 Medium $1-2
KENL7 Claude + Perplexity 🟡 Medium $5-10
KENL8 Qwen (local) 🟢 High $0
KENL9 Qwen (local) 🟢 High $0
KENL10 Qwen (local) 🟢 High $0
KENL11 Perplexity + Claude 🟡 Medium $5-10

Total cost (power user): ~$20-30/month Total cost (privacy-first): $0 (100% local AI) Total cost (balanced): ~$5-10/month (local + occasional cloud)

Next Steps

  1. Choose your AI tier (privacy-first, cost-conscious, or power user)
  2. Install Ollama (for local AI): curl -fsSL https://ollama.com/install.sh | sh
  3. Pull Qwen model: ollama pull qwen2.5:14b
  4. Get API keys (if using cloud AI):
    • Claude: console.anthropic.com
    • Perplexity: perplexity.ai/settings/api
  5. Configure modules/KENLs with your AI preferences
  6. Setup Proton services (VPN for torrenting, Mail for encrypted email, Drive for backups)

Version: 1.0.0 Last Updated: 2025-11-10 ATOM: ATOM-DOC-20251110-018