TazLab Operating Model
Level 2 (Topic) — How repositories fit together in daily work.
Concept
TazLab is not a single project — it is a set of repositories that each own a specific layer. Understanding how they fit together is essential for correct operation.
Daily Workflow
Operator shell → TazPod (tazpod/)
├── Infrastructure work → ephemeral-castle/
├── GitOps changes → tazlab-k8s/
├── Blog writing → blog-src/
├── Wiki editing → wiki.tazlab.net/
├── Agent/skill governance → SKILLS/ (via /workspace/AGENTS.md)
└── All pushed via GitHub token
Repository Roles
| Repository | Layer | Bootstrap Trigger | Steady State |
|---|---|---|---|
tazpod/ | Operator environment | n/a | tazpod up → container |
hermes-agent/ | Hermes Agent VM deployment | create.sh | VM 501 decommissioned 2026-09-26; data disk vm-501-data-orphan preserved, create.sh recreates with zero re-onboarding |
ephemeral-castle/ | Infrastructure provisioning | create.sh | Terragrunt layers |
tazlab-k8s/ | GitOps desired state | Flux bootstrap | Continuous reconciliation |
blog-src/ | Blog source | Git push | GitHub Action → Flux |
wiki.tazlab.net/ | Wiki source | Git push | GitHub Action → Flux |
SKILLS/ | OMP skills governance (loader: AGENTS.md) | n/a | Memory system |
mnemosyne-mcp-server/ | Semantic memory | Git push | GitHub Action → Flux |
Data Flow
Operator (TazPod)
│
├── Secrets → TazPod vault → S3
├── Cluster create → ephemeral-castle → Talos → K8s → Flux
├── GitOps changes → tazlab-k8s → Flux reconcile
├── Semantic memory → mnemosyne-mcp-server
└── Documentation → wiki.tazlab.net → blog.tazlab.net
See Also
- System Map: TazLab System Map
- Hub: Skills System