Skip to content

Operational Playbooks

Step-by-step execution guides for recurring workflows. Each playbook is self-contained: it lists prerequisites, the exact sequence of steps, the tools involved, and how to verify the outcome.

Choose by Goal

If you want to… Use this playbook
Transfer models and ZIMs to air-gapped servers Air-gapped Provisioning
Protect critical data with a 3-2-1 backup strategy Backup & Disaster Recovery
Validate AI-generated SQL queries for safety and performance Data Copilot SQL Validation
Automate a coding or review task end-to-end with AI assistance Dev Workflow (AI-Assisted)
Prepare office documents for LLM training or retrieval ingestion Document Preparation for LLM Training
Route emails into calendar events automatically Email to Calendar
Automate household paperwork — school letters, appointments, admin Family Admin Automation
Deploy a private, air-gapped AI stack locally Fully Offline Assistant
Configure automatic cloud-to-local failover Graceful Degradation
Deploy a 3-node HA cluster for resilient workloads K3s Cluster Setup
Audit, repair, or improve the knowledge base and documentation Knowledge Base Health
Configure NFS CSI driver for persistent volumes on K3s NFS CSI Setup
Transcribe audio and extract tasks locally Offline Transcription Pipeline
Display dashboards or a kiosk on a Raspberry Pi without a desktop Raspberry Pi Kiosk Automation
Convert a scanned document into an actionable task Scan to Task
Ingest and process school admin emails, letters, or attachments School Admin Intake
Migrate from Tailscale to a self-hosted Headscale instance Tailscale to Headscale Migration

All Playbooks

Air-gapped Provisioning

Workflow for securely transferring and verifying model weights (Ollama, vLLM, GGUF), vector datasets, and Kiwix ZIM archives onto disconnected or restricted networks.

Stack: Ollama · vLLM · Kiwix · Docker · sha256sum · FastMCP 3.1

Backup & Disaster Recovery

Comprehensive operational strategy for protecting homelab data and AI knowledge stores using a 3-2-1 backup topology with deduplicated, encrypted snapshots.

Stack: restic · BorgBackup · Kopia · MinIO / S3 · FastMCP 3.1

Data Copilot SQL Validation

A technical blueprint and operational framework for validating AI-generated SQL queries before they reach production databases. Includes static analysis, dry-runs, AST mutation, and LLM-based semantic verification.

Stack: SQLGlot · Pydantic v2 · SQLite · PostgreSQL · Qwen 3.8 / Claude 5.1

Dev Workflow (AI-Assisted)

Uses Claude Code, GitHub Actions, and local open-weight LLMs to accelerate development workflows — from issue triage to automated pull request creation and verification.

Stack: Claude Code · Ollama · GitHub Actions · n8n · FastMCP 3.1

Document Preparation for LLM Training

Normalizes docx, pdf, pptx, spreadsheets, and Google Workspace exports into machine-readable text plus metadata. Includes OCRmyPDF, Docling MCP, deduplication, and chunking guidance.

Stack: OCRmyPDF · Apache Tika · Docling MCP · Google Workspace CLI · Paperless-ngx

Email to Calendar

Parses incoming emails to extract event details and creates calendar entries automatically via n8n and CalDAV. Handles multi-lingual extraction and ambiguous date/time formats.

Stack: n8n · Radicale (CalDAV) · Ollama · Paperless-ngx · Pydantic v2

Family Admin Automation

Reduces manual overhead for household administration: school letters, medical records, appointments, shared task lists, and recurring paperwork are processed and routed automatically.

Stack: Paperless-ngx · n8n · Ollama · Nextcloud · Vikunja

Fully Offline Assistant

Operates a completely private, air-gapped AI stack on local hardware. Integrates local inference engines, web interfaces, vector search, and offline knowledge bases.

Stack: Ollama · Open WebUI · Qdrant · Kiwix · FastMCP 3.1

Graceful Degradation

Operationalizes cloud-to-local failover for LLM services during internet outages, provider rate-limiting, or cloud degradation events.

Stack: LiteLLM · Open WebUI · Ollama · vLLM · FastMCP 3.1

K3s Cluster Setup

Step-by-step operational guide for deploying a lightweight, 3-node highly available Kubernetes cluster using K3s with embedded etcd and automated ingress/DNS synchronization.

Stack: K3s · Ubuntu/Talos OS · etcd · External-DNS · Traefik

Knowledge Base Health

Audits document structure, catalog consistency, internal links, and metadata freshness. Generates a prioritized fix list and opens automated Jules remediation issues.

Stack: Python scripts · Jules · GitHub Actions · MkDocs Material

NFS CSI Setup

Configures the NFS CSI driver on a K3s cluster to connect persistent volumes to TrueNAS SCALE or dedicated ZFS storage pools with dynamic provisioning verification.

Stack: K3s · TrueNAS SCALE · Helm · NFS CSI driver

Offline Transcription Pipeline

Privacy-first workflow for converting audio and video recordings to formatted transcripts, summaries, and actionable tasks without cloud service exposure.

Stack: faster-whisper · Paperless-ngx · Vikunja · n8n · Pydantic v2

Raspberry Pi Kiosk Automation

Configures a Raspberry Pi to display rotating dashboards (Home Assistant, Grocy, calendar) in kiosk mode without a full desktop environment, featuring auto-recovery and remote management.

Stack: Raspberry Pi OS Lite · Chromium · Home Assistant · Tailscale / Headscale

Scan to Task

Scans physical documents, executes OCR, extracts structured actionable items via LLM, and creates prioritized tasks in Vikunja with low end-to-end latency.

Stack: Paperless-ngx · Apache Tika · Ollama · Vikunja · n8n

School Admin Intake

Monitors email and document repositories for school-related communications, classifies items (event / task / information), extracts key dates, and syncs to calendar and task lists.

Stack: n8n · Ollama · Radicale · Vikunja · Pydantic v2

Tailscale to Headscale Migration

Outlines the steps required to migrate homelab nodes from Tailscale SaaS servers to a self-hosted Headscale instance for complete network sovereignty and privacy.

Stack: Headscale · Tailscale · Authentik · Docker

Sources / References

Contribution Metadata

  • Last reviewed: 2027-01-07
  • Confidence: high