Skip to content

PLUGIN.md

Latest commit

 

History

History
171 lines (136 loc) · 6.32 KB

PLUGIN.md

File metadata and controls

171 lines (136 loc) · 6.32 KB
Raw
name googlecloud-plugin
description Unified Google Cloud plugin. Installs a full operational GCP team: vendor-agnostic solution designer, GCP architect, security enforcer, SRE/ops, and QA — plus 11 service skills and MCP server configurations. Design-first, security-first, Well-Architected aligned.
version 0.1.0

GoogleCloud Plugin

One plugin. Full Google Cloud capability.

Installs 5 personas + 11 service skills + MCP server configurations. All work flows through sequential quality gates. Design before code. Security before implementation. Evidence before release.

Routing Table

Route agent prompts to the correct skill using the trigger list below. When in doubt, start with solution-designer for new work or gcp-architect for GCP-scoped design.

Tier 1 — Solution Authority

Trigger Pattern Skill
Multi-vendor, multi-cloud, which-cloud, cross-cloud, overarching design skills/solution-designer/
Vendor comparison, objective architecture, hybrid cloud skills/solution-designer/

Tier 2 — GCP Domain Architecture

Trigger Pattern Skill
GCP HLD, GCP design, architect on GCP, GCP ADR skills/gcp-architect/
Agentic system, multi-agent, ADK, Agent Runtime, A2A/AP2/MCP, agent eval skills/agent-architect/
Well-Architected Framework review, GCP pillars skills/well-architected/
Terraform on GCP, IaC GCP, CFT, landing zone skills/terraform-gcp/

Tier 3 — Delivery + Assurance

Trigger Pattern Skill
Security review, IAM review, secrets, least-privilege skills/gcp-security/
Monitoring, alerting, SLO, runbook, observability skills/gcp-ops/
QA, validate, lint, smoke test, release check skills/gcp-qa/

Service Skills

Trigger Pattern Skill
Cloud Run, serverless container, deploy container skills/cloud-run/
GKE, Kubernetes, GKE cluster, node pool skills/gke/
IAM policy, service account, roles, grant access skills/iam/
BigQuery, BQ query, dataset, analytics GCP skills/bigquery/
Cloud Storage, GCS, bucket, object storage skills/cloud-storage/
Vertex AI, Gemini GCP, model, endpoint, agent builder skills/vertex-ai/
VPC, firewall, load balancer, networking GCP skills/networking/
Cloud Logging, Cloud Monitoring, alerting, trace skills/logging-monitoring/

Cross-Cutting

Trigger Pattern Skill
MCP server, MCP setup, GenAI Toolbox skills/mcp-servers/

Gate System (Sequential — Non-Negotiable)

New work or feature request
    ↓
[SOLUTION GATE] solution-designer
    Skip if: unambiguously GCP-only, no cross-vendor components
    ↓
[GCP DESIGN GATE] gcp-architect (+ agent-architect if agentic)
    Block if: no GCP HLD, or HLD contradicts solution design
    Agentic workloads: agent-architect defines topology, protocols, model, eval set
    ↓
[SECURITY GATE] gcp-security
    Block if: IAM over-permissive, secrets in code, GCP risks unaddressed
    ↓
[IMPLEMENTATION] service skills
    Warn before every billable action
    ↓
[QUALITY GATE] gcp-qa  →  make gate  (+ agent-architect agent-eval if agentic)
    Block if: validate/lint/test fails, acceptance criteria unmet
    Agentic: agent eval (groundedness, task success, trajectory) must pass
    ↓
[OPERATIONAL READINESS GATE] gcp-ops
    Block if: no SLO, no alerting, no runbook for production workload
    ↓
Release

Validation (3-Minute Gate)

make gate       # validate + lint + test — must pass before any merge to main
make validate   # SKILL.md frontmatter contract
make lint       # reference URL liveness
make test       # pytest smoke tests
make check      # freshness: hash drift detection
make crawl      # research pipeline (weekly)

Plugin Structure

googlecloud-plugin/
├── PLUGIN.md               # This file — top-level orchestrator
├── plugin.yaml             # Metadata, version, full skill index
├── Makefile                # Validation targets (make gate)
├── skills/                 # All skill folders
│   ├── solution-designer/  # Tier 1: vendor-agnostic
│   ├── gcp-architect/      # Tier 2: GCP design authority
│   ├── gcp-security/       # Tier 3: security gate
│   ├── gcp-ops/            # Tier 3: SRE / ops
│   ├── gcp-qa/             # Tier 3: quality gate
│   ├── cloud-run/          # Service skill
│   ├── gke/                # Service skill
│   ├── iam/                # Service skill
│   ├── bigquery/           # Service skill
│   ├── cloud-storage/      # Service skill
│   ├── vertex-ai/          # Service skill
│   ├── networking/         # Service skill
│   ├── logging-monitoring/ # Service skill
│   ├── mcp-servers/        # MCP configuration + setup
│   ├── well-architected/   # GCP WAF reference
│   └── terraform-gcp/      # IaC patterns
├── architecture/
│   ├── HLD/                # High-level designs
│   ├── LLD/                # Low-level designs
│   ├── decisions/          # ADRs (ADR-001 onward)
│   └── patterns/           # Reusable GCP patterns
├── engineering/
│   ├── features/           # Feature implementation docs
│   ├── use-cases/          # Business use case definitions
│   └── runbooks/           # Operational runbooks
├── shared/
│   ├── auth/               # ADC + SA patterns
│   └── discovery/          # Agent skill enumeration
├── research/               # Sources + coverage matrix
├── scripts/                # Research + validation scripts
└── tests/                  # Smoke tests

Auth (Shared)

All skills use Application Default Credentials (ADC) unless the skill specifies otherwise.

# Developer
gcloud auth application-default login

# CI/CD (impersonate SA — no key file)
gcloud auth application-default login \
  --impersonate-service-account=SA@PROJECT.iam.gserviceaccount.com

See shared/auth/ for full patterns.

Coverage Matrix

See research/coverage-matrix.md for service-by-service skill status.

Target for v0.1: zero missing entries for all first-wave services.