GCP-to-AWS Migration Skill

Philosophy

Re-platform by default: Select AWS services that match GCP workload types (e.g., Cloud Run → Fargate, Cloud SQL → RDS).
Dev sizing unless specified: Default to development-tier capacity (e.g., db.t4g.micro, single AZ). Upgrade only on user direction.
Infrastructure-first approach: v1.0 migrates Terraform-defined infrastructure only. App code scanning and billing import are v1.1+.

Prerequisites

User must provide GCP infrastructure-as-code:

One or more .tf files (Terraform)
Optional: .tfvars or .tfstate files

If no Terraform files are found, stop immediately and ask user to provide them.

State Management

Migration state lives in .migration/[MMDD-HHMM]/ directory (created by Phase 1, persists across invocations):

.migration/
├── .gitignore                       # Auto-created to protect state files from git
└── 0226-1430/                       # MMDD-HHMM timestamp
    ├── .phase-status.json           # Current phase tracking
    ├── gcp-resource-inventory.json  # All GCP resources found
    ├── gcp-resource-clusters.json   # Clustered resources by affinity
    ├── clarified.json               # User answers (Phase 2 output)
    ├── aws-design.json              # AWS services mapping (Phase 3 output)
    ├── estimation.json              # Cost breakdown (Phase 4 output)
    └── execution.json               # Timeline + risks (Phase 5 output)

Note: The .migration/ directory is automatically protected by a .gitignore file created in Phase 1. Migration state files will not be accidentally committed to version control.

.phase-status.json schema:

{
  "phase": "discover|clarify|design|estimate|execute",
  "status": "in-progress|completed",
  "timestamp": "2026-02-26T14:30:00Z",
  "version": "1.0.0"
}

If .phase-status.json exists:

If status is completed: advance to next phase (discover→clarify, clarify→design, etc.)
If status is in-progress: resume from that phase

Phase Routing

On skill invocation: Check for .migration/*/ directory
- If none exist: Initialize Phase 1 (Discover), set status to in-progress
- If multiple exist: STOP. Output: "Multiple migration sessions detected in .migration/:" then for each directory, display its name and the contents of its .phase-status.json (phase + status). Output: "Pick one to continue: [list with phase info]"
- If exists: Load .phase-status.json and validate:
  - If empty file (0 bytes): STOP. Output: "State file is empty. Delete .migration/[MMDD-HHMM]/.phase-status.json and restart."
  - If invalid JSON: STOP. Output: "State file corrupted (invalid JSON). Delete .migration/[MMDD-HHMM]/.phase-status.json and restart Phase [X]."
  - If missing required fields (phase, status, timestamp, version): STOP. Output: "State file incomplete (missing [field]). Delete and restart."
  - If version != "1.0.0": STOP. Output: "Incompatible state file version: [version]. This skill requires version 1.0.0."
  - If unrecognized phase value: STOP. Output: "Unrecognized phase: [value]. Valid values: discover, clarify, design, estimate, execute."
  - If status not in {in-progress, completed}: STOP. Output: "Unrecognized status: [value]. Valid values: in-progress, completed."
  - If valid: Determine next action:
    - If phase status is in-progress: Resume that phase
    - If phase status is completed: Advance to next phase
Phase transition mapping (when phase is completed):
- discover (completed) → Route to clarify
- clarify (completed) → Route to design
- design (completed) → Route to estimate
- estimate (completed) → Route to execute
- execute (completed) → Migration complete; offer summary and cleanup options
Phase gate checks: If prior phase incomplete, do not advance (e.g., cannot enter estimate without completed design)

Phase Summary Table

Phase	Inputs	Outputs	Reference
Discover	`.tf` files	`gcp-resource-inventory.json`, `gcp-resource-clusters.json`, `.phase-status.json` updated	`references/phases/discover/discover.md`
Clarify	`gcp-resource-inventory.json`, `gcp-resource-clusters.json`	`clarified.json`, `.phase-status.json` updated	`references/phases/clarify.md`
Design	`gcp-resource-inventory.json`, `gcp-resource-clusters.json`, `clarified.json`	`aws-design.json`, `aws-design-report.md`, `.phase-status.json` updated	`references/phases/design.md`
Estimate	`aws-design.json`, `clarified.json`	`estimation.json`, `estimation-report.md`, `.phase-status.json` updated	`references/phases/estimate.md`
Execute	`aws-design.json`, `estimation.json`	`execution.json`, `execution-timeline.md`, `.phase-status.json` updated	`references/phases/execute.md`

MCP Servers

awspricing (for cost estimation):

Call get_pricing_service_codes() to detect availability
If success: use live AWS pricing
If timeout/error: fall back to references/shared/pricing-fallback.json (includes 2026 on-demand rates for major services)

awsknowledge (for design validation):

Use for regional availability checks (e.g., service available in target region?)
Use for feature parity checks (e.g., do required features exist in AWS service?)
Use for service constraints and best practices
Fallback: if unavailable, set validation_status: "skipped" in aws-design.json with note in design report
Important: Validation is informational; design proceeds either way (not blocking)

Error Handling

Condition	Action
No `.tf` files found	Stop. Output: "No Terraform files detected. Please provide `.tf` files with your GCP resources and try again."
`.phase-status.json` missing phase gate	Stop. Output: "Cannot enter Phase X: Phase Y-1 not completed. Start from Phase Y or resume Phase Y-1."
awspricing unavailable after 3 attempts	Display user warning about ±15-25% accuracy. Use `pricing-fallback.json`. Add `pricing_source: fallback` to estimation.json.
User does not answer all Q1-8	Offer Mode C (defaults) or Mode D (free text). Phase 2 completes either way.
`aws-design.json` missing required clusters	Stop Phase 4. Output: "Re-run Phase 3 to generate missing cluster designs."

Defaults

IaC output: None (v1.0 produces design, cost estimates, and execution plans — no IaC code generation)
Region: us-east-1 (unless user specifies, or GCP region → AWS region mapping suggests otherwise)
Sizing: Development tier (e.g., Aurora Serverless v2 0.5 ACU for databases, 0.5 CPU for Fargate)
Migration mode: Full infrastructure path (no AI-only subset path in v1.0)
Cost currency: USD
Timeline assumption: 8-12 weeks total

Workflow Execution

When invoked, the agent MUST follow this exact sequence:

Load phase status: Read .phase-status.json from .migration/*/.
- If missing: Initialize for Phase 1 (Discover)
- If exists: Determine current phase based on phase field and status value
Determine phase to execute:
- If status is in-progress: Resume that phase (read corresponding reference file)
- If status is completed: Advance to next phase (read next reference file)
- Phase mapping for advancement:
  - discover (completed) → Execute clarify (read references/phases/clarify.md)
  - clarify (completed) → Execute design (read references/phases/design.md)
  - design (completed) → Execute estimate (read references/phases/estimate.md)
  - estimate (completed) → Execute execute (read references/phases/execute.md)
  - execute (completed) → Migration complete
Read phase reference: Load the full reference file for the target phase.
Execute ALL steps in order: Follow every numbered step in the reference file. Do not skip, optimize, or deviate.
Validate outputs: Confirm all required output files exist with correct schema before proceeding.
Update phase status: Each phase reference file specifies the final .phase-status.json update (records the phase that just completed).
Display summary: Show user what was accomplished, highlight next phase, or confirm migration completion.

Critical constraint: Agent must strictly adhere to the reference file's workflow. If unable to complete a step, stop and report the exact step that failed.

User can invoke the skill again to resume from last completed phase.

Scope Notes

v1.0 includes:

Terraform infrastructure discovery (no app code scanning)
User requirement clarification (8 structured questions)
Structured Design (cluster-based mapping from Terraform)
AWS cost estimation (from pricing API or fallback)
Execution timeline and risk assessment

Deferred to v1.1+:

App code scanning (runtime detection of compute workload types)
AI-only fast-track path in Clarify/Design
Billing data import from GCP
Flat Design path (for non-Terraform codebases)

gcp-to-aws