Platform Engineer Guide
Design and govern reusable modules (golden paths) with Argy Studio, approvals, and observability.
This guide focuses on the Platform Engineer workflow in Argy: designing reusable modules, publishing golden paths, and enforcing governance without blocking teams.
Module Studio (Visual workflow editor)
The Studio is the central interface to design automation modules using a visual canvas.
Canvas
- Nodes: Start, End, and intermediate actions
- Edges: directed connections between nodes (DAG)
- Zoom/pan, grid alignment, multi-select
Action palette (50+)
Examples of action categories:
- Cloud & IaC (Kubernetes, VPC, databases, storage, DNS)
- CI/CD (GitHub Actions, GitLab CI, Jenkins, Helm deploy)
- Security (Trivy, SonarQube, OWASP ZAP, Gitleaks, Kyverno policy)
- Observability (Prometheus, Grafana, PagerDuty, k6)
- Notifications (Slack, Teams, Email, Webhooks)
- Governance (compliance checks, policy validation, audit log)
- AI agent actions (LLM-driven execution with prompts and tools)
Node configuration
Each node supports:
- Inputs: JSON schema (types, validation, defaults)
- Outputs: schema of produced data
- Bindings: reference other node outputs via
{{nodeLabel.outputs.key}}
Built-in assistant
The Studio assistant can:
- Chat to build complete workflows from natural language
- Modify workflows incrementally (add/remove nodes)
- Configure inputs/outputs/bindings
- Simulate workflows
- Publish modules
Argy AI action (custom subagent)
Modules can include an Argy AI action: an LLM-driven step with custom prompts and tools. It can orchestrate sub-agents as part of a workflow, enabling companies to build their own governed AI agents inside golden paths.
Simulation
Before publishing, modules can be simulated:
- Sandbox execution in the orchestrator
- Real-time logs (WebSocket)
- Structured outputs
- Automatic validation (cycles, missing connections, schema constraints)
Publishing & versioning
- Semantic versioning (auto-increment patch)
- Changelog per version
- Visibility: private, organization, or public
- Automatic thumbnail generation
Golden Paths (module catalog)
Golden Paths are versioned, validated, and approved modules that encode best practices.
Lifecycle
Draft (Studio) → Published (Catalog) → Deployed (Project) → Deprecated
Status badges
- Stable
- Experimental
- Deprecated
- Draft
Module spec (v1)
Modules are represented as a native JSON spec (inputs/outputs schemas + workflow graph).
{
"name": "microservice-secure-deploy",
"version": "2.1.0",
"description": "Deploy a secure microservice with Trivy scan",
"inputsSchema": { "type": "object" },
"outputsSchema": { "type": "object" },
"workflow": { "nodes": [], "edges": [] }
}
Composition (module-in-module)
Published modules automatically appear as actions in the Studio palette, enabling composition.
Projects, deployments, and pipelines
Projects
A project represents an application or service with:
- Metadata
- A module (golden path) as the baseline
- Environments (Dev/Staging/Prod)
- Deployment history and metrics
Deployment wizard
Deployments follow a 3-step flow:
- Select module and version
- Configure inputs (forms generated from JSON schema)
- Review and create
Run statuses
QUEUED,WAITING_APPROVAL,RUNNING,SUCCEEDED,FAILED,CANCELED
Pipeline views
- Global runs timeline
- By project / module / environment
- Real-time logs and outputs
- Actions: cancel, retry, approve/reject, export logs
Governance, approvals, and audit
- Approval policies can require 1 to 10 approvals for actions/resources (modules, projects, deployments, policies).
- Decisions are notified (email/Slack/Teams) and fully audited.
- Audit logs are exportable (CSV) with a minimum retention of 90 days.
Integrations
Argy integrates with:
- Git: GitHub, GitLab, Bitbucket, Azure DevOps
- CI/CD: GitHub Actions, GitLab CI, Jenkins, Azure DevOps Pipelines
- Cloud: AWS, Azure, GCP
- Secrets: Azure Key Vault, HashiCorp Vault, AWS Secrets Manager
- Observability: Prometheus, Grafana, Datadog, Azure Monitor, New Relic
- Identity: Entra ID, Okta, Google Workspace (OIDC/SAML) + SCIM