Concept · Spec complete

Know before you deploy.

A complete concept for deterministic and AI-assisted Kubernetes validation: scheduling simulation, capacity reasoning, and CI checks before deployment.

Discuss the concept Read the architecture

The kubectl gap

kubectl apply --dry-run Kubernetes dry-run can validate requests without persisting them, but does not demonstrate that a future workload will fit the cluster. PRELUDE explores that separate scheduling and capacity question.

Will the deployment fit? Are there enough resources? Will it trigger autoscaling? Will pod anti-affinity rules prevent scheduling? Will it evict existing workloads?

PRELUDE proposes a way to explore these questions before deployment. It is a specification, not an implemented validator.

Three proposed validation layers

Deterministic Engine

The proposed deterministic engine would parse manifests, calculate requests and limits, and simulate scheduling constraints: node affinity, topology spread, and PVC binding. No implementation is released.

AI Reasoning

The specification proposes Architecture and Capacity agents for pattern analysis, scaling scenarios, and cost estimates, with A2A communication. These are design targets, not validated predictions.

Integration

Proposed interfaces include a REST API, CLI, ArgoCD PreSync hook, and MCP server. None is offered as a released integration.

How it fits together

This diagram represents the specification: a proposed deterministic foundation, optional AI reasoning, and planned CI/CD interfaces. It does not depict a deployed system.

Illustrative CLI specification

The commands and results below are fictional examples from the design specification. There is no installable PRELUDE CLI and these outputs are not test results.

terminal

# Specification example — not an executable CLI or a test result

# Validate a deployment against your cluster
prelude validate deployment.yaml --cluster prod-gke

# Output
✓ Manifest parsed (Deployment: api-server, 3 replicas)
✓ Resource requirements: 1.5 CPU, 3Gi memory (total)
✓ Scheduling simulation: all 3 replicas schedulable
✓ Node pool: e2-standard-4 has capacity (62% utilized → 74%)
⚠ Warning: No PodDisruptionBudget defined
⚠ Warning: Memory request/limit ratio is 1:4 (burst risk)

# With AI reasoning enabled
prelude validate deployment.yaml --cluster prod-gke --ai

✓ Deterministic validation passed
✓ AI Architecture Agent: deployment pattern is sound
⚠ AI Capacity Agent: at current growth rate (12%/mo),
  node pool will need scaling in ~18 days
  Recommendation: increase node pool max from 5 to 8

Proposed capabilities

Ten design targets across three proposed layers. No capabilities in this list have been released.

Manifest parsing

Deterministic

Deployments, StatefulSets, DaemonSets, Jobs, CronJobs.

Resource calculation

Deterministic

CPU, memory, GPU, ephemeral storage across all replicas.

Scheduling simulation

Deterministic

Node affinity, anti-affinity, taints, tolerations, topology spread.

PVC binding validation

Deterministic

StorageClass availability, capacity, access modes.

Conflict detection

Deterministic

Port conflicts, resource name collisions, quota violations.

Pattern analysis

Deployment anti-patterns, best practice recommendations.

Capacity prediction

Growth modeling, scaling timeline, cost estimation.

Risk assessment

Eviction probability, OOM risk, scheduling failure scenarios.

CI/CD integration

Integration

REST API, CLI exit codes, ArgoCD PreSync hook.

MCP server

Integration

AI-assisted operations through Model Context Protocol.

Scope of the specification

Planned validation areas, not implemented functionality or a verified comparison with other tools.

	PRELUDE
Syntax validation	Proposed — not implemented
Policy validation	Proposed — not implemented
Capacity simulation	Proposed — not implemented
Scheduling simulation	Proposed — not implemented
AI recommendations	Proposed — not implemented
Cost prediction	Proposed — not implemented
CI/CD integration	Proposed — not implemented
Open-source	Apache-2.0 planned

Proposed tech stack

Technology choices documented in the specification; not evidence of an implemented product.

Technology	Role
Python	Core runtime
FastAPI	REST API
LangGraph	AI agent orchestration
A2A Protocol	Agent-to-agent communication
MCP	Model Context Protocol integration
GKE	Primary target (Autopilot + Standard)

Where we are

PRELUDE currently exists as a complete specification and architecture, not a released product. The deterministic engine — manifest parsing, resource calculation, scheduling simulation — is the proposed foundation. Optional AI reasoning layers build on top.

Apache-2.0 is the intended license recorded in the specification. No source implementation or public repository has been released, so this is a plan rather than a current software license grant.

Explore the specification.

PRELUDE is a documented concept, not a released repository. Talk to us about the architecture or follow related work in Field Notes.

Discuss PRELUDE Field Notes