Documentation Index — `devops-infra-helm-charts`

Repo: GitOps Helm values + cached/forked charts for Meesho's GKE infrastructure fleet.

Layer: Layer 1 — Agent-Writable (config repo).

Repo entry point for agents: /CLAUDE.md.

This index mirrors the structure of the agent-documentation pyramid: governance at the top, schemas underneath, procedures and runbooks above the operational floor, skills wrapping procedures for agents, ADRs explaining the why.

Layout

/
├── CLAUDE.md                                 Agent entry point (Layer + naming + layout + layer table)
├── README.md                                 Short repo intro for humans
├── index.md                                  ← you are here
├── repository.yaml                           Owner metadata (registry-bootstrap-managed)
│
├── docs/
│   ├── architecture.md                       Full deploy lifecycle, fleet, chart inventory, hooks
│   ├── global/                               Cross-cutting governance
│   │   ├── AGENT_BOUNDARIES.md               Layer 1 / 2 / 3 map for THIS repo, blast radii
│   │   ├── SANCTITY_RULES.md                 Numbered non-negotiable rules
│   │   └── coding-guidelines/
│   │       └── helm-values.md                Patterns for authoring custom-values.yaml
│   └── platform/
│       ├── procedures/                       "How to make change X" — six recipes
│       │   ├── onboard-app-to-cluster.md
│       │   ├── onboard-new-cluster.md
│       │   ├── update-chart-version.md
│       │   ├── fork-upstream-chart.md
│       │   ├── blue-green-chart-migration.md
│       │   └── deboard-app.md
│       ├── runbooks/                         Symptom → diagnosis decision trees
│       │   ├── argocd-sync-failure.md
│       │   ├── ingress-down.md
│       │   └── pod-pending-scheduling.md
│       └── schemas/                          Field-by-field YAML annotation
│           ├── custom-values-schema.md
│           ├── raw-manifest-sidecar-schema.md
│           └── storageclass-priorityclass-schema.md
│
├── skills/
│   └── infra/                                Agent-callable parameterised tasks
│       ├── onboard-app.md
│       ├── bump-chart-version.md
│       └── diagnose-scheduling.md
│
├── wiki/
│   ├── entities/
│   │   └── DevOps Infra Helm Charts.md       Architectural reference entity for THIS repo
│   └── analyses/                             Architecture Decision Records
│       ├── ADR-A1-cache-vs-upstream-charts.md
│       ├── ADR-A2-blue-green-sibling-pattern.md
│       ├── ADR-A3-per-cluster-scheduling.md
│       ├── ADR-A4-raw-manifest-sidecars-in-helm-overrides.md
│       └── ADR-A5-manual-sync-default-for-infra.md
│
├── contour-nodeselector-tolerations-summary.md
│                                             Per-cluster Contour scheduling matrix
│                                             (load-bearing — read before any Contour edit)
│
├── helm-templates/<chart>/                   74 cached/forked upstream charts
├── helm-overrides/<cluster>/<app>/           Cluster × app override values + sidecar manifests
├── manifests/                                Cluster-wide singletons (storageclass / priorityclass / Jenkins / JFrog)
├── pre-commit-scripts/                       TruffleHog (active); CAC + Yaak (no-op here)
└── post-commit-scripts/                      Cursor metric collector (background, non-blocking)

"I'm an agent. What do I read first?"

/CLAUDE.md — entry point. Repo role, layer, NEVER DO list, layout, layer constraint table.
docs/global/AGENT_BOUNDARIES.md — per-operation Layer 1 / 2 / 3 classification with blast radii.
docs/global/SANCTITY_RULES.md — numbered hard stops.
The relevant procedure or runbook for the task.

"I'm a human reviewing an agent-generated PR. What do I check?"

docs/platform/schemas/custom-values-schema.md — does each field follow convention?
docs/global/SANCTITY_RULES.md — does the PR cross any non-negotiable?
The procedure used (if any) — did the PR follow it end-to-end?
contour-nodeselector-tolerations-summary.md — if Contour is touched, do nodeSelector/tolerations match the cluster's row?

"Why is this repo shaped this way?"

Read in order:

docs/architecture.md — the system context, module boundaries, deploy lifecycle.
ADRs in wiki/analyses/ — five recorded decisions.

"Help me do task X."

Task	Doc
Add a new app override to a cluster	docs/platform/procedures/onboard-app-to-cluster.md
Bring up a brand-new cluster's overrides	docs/platform/procedures/onboard-new-cluster.md
Bump a chart's pinned dependency version	docs/platform/procedures/update-chart-version.md
Intentionally fork an upstream chart	docs/platform/procedures/fork-upstream-chart.md
Migrate a chart blue-green (sibling pattern)	docs/platform/procedures/blue-green-chart-migration.md
Remove a retired app override	docs/platform/procedures/deboard-app.md
Argo CD app errored / OutOfSync	docs/platform/runbooks/argocd-sync-failure.md
Ingress (Contour) is down on a cluster	docs/platform/runbooks/ingress-down.md
Pods pending / wrong-node scheduling	docs/platform/runbooks/pod-pending-scheduling.md

Repo facts (quick reference)


Charts in `helm-templates/`	74
Cluster directories under `helm-overrides/`	30+ (`k8s--prd-ase1[c]`, `k8s-shared-int-ase1`, `k8s-aurva-prd-ase1`, `k8s-supply-dev-ase1`, plus `db-` dataplane)
Singletons under `manifests/`	StorageClasses (4), per-cluster PriorityClasses, Jenkins/JFrog filestore PV/PVCs
Active pre-commit hooks	TruffleHog (verified-secret scan; webhook to `observe.meeshogcp.in`)
Build / test / lint	None — declarative YAML only
Deploy mechanism	Argo CD reconcile from `main`
Owners	`siddharth.pal@meesho.com` (primary), `samarth.nag@meesho.com` (secondary) — per `repository.yaml`

6.9 KiB Raw Blame History Unescape Escape

Documentation Index — devops-infra-helm-charts