Repository Map & Taxonomy¶

The Edge-Computing-LLM ecosystem consists of 8 dedicated GitHub repositories under the Edge-Computing-LLM organization. Each repository owns a specific operational domain, ensuring clean separation of concerns, zero code duplication, and strict runtime boundaries.

Ecosystem Repository Taxonomy¶

Repository Name	Role / Domain	Primary Deliverables & Artifacts	Out of Scope / Does Not Own
`.github`	Organization Governance	Org profile, issue templates, SECURITY.md, CONTRIBUTING.md, license policies	Code execution, deployment logic
`edge-cli`	Unified Control Plane	`edge` Go binary, CLI config (`edge-cli.yaml`), dry-run engine, dependency gates	Helm chart definitions, model binaries
`k3s-nvidia-edge`	Layer 1 Substrate	GPU Operator Helm chart, Container Toolkit config, `RuntimeClass/nvidia`, `pkg/edgebase`	Ollama, model files, observability dashboards
`llm-observability-stack`	Layer 2 Workloads	Helm chart, model profiles (`values.*.yaml`), Grafana dashboards, Ollama, Open WebUI, OTel	Host drivers, CUDA installation
`edge-llm-tests`	Cross-Repo Verifier	Go test suite (`edge-llm-tests`), benchmark harness, sanitized JSON output	Mutating cluster state, installing packages
`gguf-observability`	GGUF Runtime Contract	`gguf-observe` binary, GGUF memory ceiling verifier, model contract profiles	Weight downloading, pod lifecycle
`Edge-Computing-LLM-Documentation`	Central Docs Site	Zensical documentation site, architecture diagrams, operational runbooks	Duplicating Go source or Helm templates
`edge-computing-openai`	EdgeProof Verifier & Codex	Codex Agent Skill, `edge-openai` verifier CLI, Build Week verification evidence	Unapproved cluster mutations or repairs

Local Project Directory Mapping¶

When developing or managing the platform locally on /media/waqasm86/External1/Waqas-Projects/Project-Linux-Kubernetes-Nvidia/Project-Edge-Computing-LLM/, the directory layout maps directly to the organization repositories:

Project-Edge-Computing-LLM/
├── edge-cli/                            --> https://github.com/Edge-Computing-LLM/edge-cli
├── k3s-nvidia-edge/                     --> https://github.com/Edge-Computing-LLM/k3s-nvidia-edge
├── llm-observability-stack/             --> https://github.com/Edge-Computing-LLM/llm-observability-stack
├── gguf-observability/                  --> https://github.com/Edge-Computing-LLM/gguf-observability
├── edge-llm-tests/                      --> https://github.com/Edge-Computing-LLM/edge-llm-tests
├── edge-computing-openai/               --> https://github.com/Edge-Computing-LLM/edge-computing-openai
├── Edge-Computing-LLM-Documentation/   --> https://github.com/Edge-Computing-LLM/Edge-Computing-LLM-Documentation
├── dashboard-template-candidates/       --> Grafana dashboard reference templates
├── Credentials-For-LLM-Observability/   --> Encrypted/sanitized credentials reference
├── Project-Notes/                       --> Development logs and operational notes
└── _analysis/                           --> Point-in-time cluster analysis snapshots

Upstream Project Mapping¶

The platform builds upon industry-standard cloud-native and edge AI projects:

mindmap
  root((Edge-Computing-LLM))
    Substrate Runtimes
      k3s (Lightweight Kubernetes)
      containerd (OCI Runtime)
      NVIDIA GPU Operator
      NVIDIA Container Toolkit
      NVIDIA DCGM Exporter
    Inference & UI
      Ollama Engine
      GGUF Quantization Runtimes
      Open WebUI
      Redis Cache
    Observability & Telemetry
      OpenTelemetry Collector & Operator
      Prometheus & Kube-State-Metrics
      Grafana & Alertmanager
    Tooling & Controls
      Go (1.25+)
      Helm (v3+)
      Zensical / MkDocs

Source-of-Truth Rule¶

While this central documentation site provides comprehensive architecture guides, step-by-step tutorials, and operational runbooks, the code, Go interfaces, and checked-in values.yaml files within each respective repository remain the ultimate source of truth.