agentleFS
Sign inStart free
NVIDIAKnown publisherNVIDIA/nvcf/deploy/stacks/observability/AGENTS.md

nvcf / observability

AGENTS.md261 starsChanged 9 days agoApache-2.0View on GitHub

This is the shared NVCF observability Helmfile stack. It owns cluster-level observability infrastructure for self-managed control-plane and compute-plane deployments. - Keep shared collector, target allocator, Prometheus Operator CRD, OpenTelemetry Operator, optional VictoriaMetrics, and default monitor wiring here. - Keep application metrics endpoints, labels, port names, and paths in the application charts that expose those endpoints. - Do not add control-plane services or compute-plane services here. - Keep the reusable base profile disabled. - Derive control-plane and compute-plane behavior from observability.profile;…

The file

We split this AGENTS.md into the rules it gives an agent

11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep shared collector, target allocator, Prometheus Operator CRD, OpenTelemetry Operator, optional VictoriaMetrics, and default monitor …

    line 11
  2. Distinctive:

    Keep application metrics endpoints, labels, port names, and paths in the application charts that expose those endpoints.

    line 14
  3. Distinctive:

    Do not add control-plane services or compute-plane services here.

    line 16
  4. Very specific:

    Keep the reusable base profile disabled.

    line 17
  5. Very specific:

    Do not use Helm lookup to silently change profiles or component ownership based on live cluster state.

    line 40
  6. Very specific:

    Run profile and integration assertions with make test.

    line 47
  7. Distinctive:

    Run git diff --check before committing.

    line 48
  8. Distinctive:

    Keep profile derivation and validation at the top of the Helmfile.

    line 52
  9. Distinctive:

    Keep chart templates data-driven.

    line 53
  10. Distinctive:

    Put stable scrape endpoints, labels, ports, and paths in the application chart that exposes them.

    line 54
  11. Very specific:

    Keep reusable defaults in environments/base.yaml.

    line 59

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md - Observability Stack

Purpose

This is the shared NVCF observability Helmfile stack. It owns cluster-level observability infrastructure for self-managed control-plane and compute-plane deployments.

Scope

  • Keep shared collector, target allocator, Prometheus Operator CRD, OpenTelemetry Operator, optional VictoriaMetrics, and default monitor wiring here.
  • Keep application metrics endpoints, labels, port names, and paths in the application charts that expose those endpoints.
  • Do not add control-plane services or compute-plane services here.
  • Keep the reusable base profile disabled.
  • Derive control-plane and compute-plane behavior from observability.profile; do not expose parallel plane booleans.
  • Enabled profiles install shared components once by default. Fine-grained component modes are install, existing, and disabled.
  • Derive the NVCA collector and BYOObservability feature gate from compute and all. Preserve explicit compute-plane overrides.

Key Files

  • helmfile.d/01-observability.yaml.gotmpl: shared observability releases
  • environments/base.yaml: scaffold defaults and scrape contracts
  • charts/nvcf-otel-collector: OpenTelemetryCollector resource and Target Allocator RBAC
  • charts/nvcf-default-monitors: centrally owned concrete monitor resources
  • values/victoria-metrics.yaml.gotmpl: values bridge for the VictoriaMetrics chart

Profiles

  • disabled: render no observability runtime resources.
  • control: install control-plane metrics and autoscaler backend defaults.
  • compute: install NVCA, DCGM, and worker metric defaults.
  • all: install the union once.

Do not use Helm lookup to silently change profiles or component ownership based on live cluster state. The consuming stack must choose the profile explicitly.

Development

  • Render the local environment with make template HELMFILE_ENV=local.
  • Run profile and integration assertions with make test.
  • Run git diff --check before committing.

Code Style

  • Keep profile derivation and validation at the top of the Helmfile.
  • Keep chart templates data-driven; use one generic template per monitor kind.
  • Put stable scrape endpoints, labels, ports, and paths in the application chart that exposes them.

Subtree Conventions

  • Keep reusable defaults in environments/base.yaml; environment files should contain only overrides.
  • Preserve explicit user values when applying profile defaults.
  • Validate every new mode, dependency, and rendered monitor shape in tests/profile-defaults.sh.
  • Bump version and appVersion in a chart's Chart.yaml when the chart changes after a stack release. tests/chart-versions.sh enforces this.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free