Provider Resources

Terraform Provider Resources Implementation Guide

Overview

This guide covers developing Terraform Provider resources and data sources. Resources represent infrastructure objects that Terraform manages through Create, Read, Update, and Delete (CRUD) operations.

Use the Plugin Framework (opens in a new tab) for all net-new resources and data sources. Plugin SDKv2 is for maintaining resources that already exist on it; do not write new code against it. A provider can serve both during migration by muxing (terraform-plugin-mux (opens in a new tab)), so adopting the Framework never requires a big-bang rewrite. To tell which mode an existing provider is in, check go.mod: terraform-plugin-mux present means it serves both SDKv2 and Framework code; only terraform-plugin-sdk/v2 means SDKv2-only; only terraform-plugin-framework means Framework-only. Be cautious about migrating existing SDKv2 resources: the Framework distinguishes null from zero values, so naive migrations change behavior for existing users (use the provider-framework-migration skill, if available).

References (load when needed):

• references/design-principles.md — what should (and should not) become a resource; data source semantics; relationship and async-task modeling
• references/retries-and-waiters.md — eventual consistency, retry patterns, and status/wait function structure

File Structure

Most providers keep every resource in a single package:

internal/provider/
├── provider.go                  # Provider schema + Configure
├── widget_resource.go           # Resource implementation
├── widget_resource_test.go      # Acceptance tests
├── widget_data_source.go        # Data source (if applicable)
└── widget_data_source_test.go

Large multi-service providers (e.g. terraform-provider-aws) split into internal/service/<service>/ packages instead, with an idiomatic file taxonomy worth adopting once a package grows: consts.go, find.go (finders), status.go (status functions), wait.go (waiters), sweep.go (test sweepers), exports_test.go.

Documentation lives in docs/ and is generated with tfplugindocs:

docs/
├── resources/<name>.md          # generated; optional <name>.md.tmpl template
└── data-sources/<name>.md

(Hand-written website/docs/r/*.html.markdown trees exist in some older, large providers — follow the target repo’s convention when editing one.)

Resource Structure

A Framework resource is a struct holding the API client, with interface assertions making the implemented behaviors explicit:

var (
    _ resource.Resource                = &widgetResource{}
    _ resource.ResourceWithConfigure   =

How the provider’s Configure produces that client — schema, credential resolution, validation — is covered by the provider-configuration skill (if available).

On id: SDKv2 required a magic id attribute; the Framework does not. If the API has its own identifier, expose it under its real meaning and do not add a second, redundant id. Only keep id when it is the API’s identifier (as above).

CRUD Operations

Create

func (r *widgetResource) Create(ctx context.Context, req resource.CreateRequest, resp *resource.CreateResponse) {
    var data widgetResourceModel
    resp.Diagnostics.

Read

Read must handle out-of-band deletion by removing the resource from state so the next plan recreates it, rather than erroring forever:

func (r *widgetResource) Read(ctx context.Context, req resource.ReadRequest, resp *resource.ReadResponse) {
    var data widgetResourceModel
    resp.Diagnostics.Append

Update

Only call the API for attributes that actually changed; compare plan against state:

func (r *widgetResource) Update(ctx context.Context, req resource.UpdateRequest, resp *resource.UpdateResponse) {
    var plan, state widgetResourceModel
    resp.Diagnostics.

Delete

Treat “already gone” as success — the desired end state is reached:

func (r *widgetResource) Delete(ctx context.Context, req resource.DeleteRequest, resp *resource.DeleteResponse) {
    var data widgetResourceModel
    resp.Diagnostics.Append(req.State.Get(ctx, 

Import

With ResourceWithImportState asserted, passthrough of the identifier is one line:

func (r *widgetResource) ImportState(ctx context.Context, req resource.ImportStateRequest, resp *resource.ImportStateResponse) {
    resource.ImportStatePassthroughID(ctx, path.Root("id"), req, resp)
}

For multi-part identifiers, parse a delimited import ID (commonly comma-separated) and set each attribute explicitly.

Resource Design Principles

Before implementing, check the shape of the thing being modeled (full treatment in references/design-principles.md):

• A resource is the smallest useful building block; if the API offers CRUD for it, it likely deserves its own resource.
• A resource should talk to one API/service only — cross-service resources break permissions, auditing, and endpoint configuration.
• Data sources are read-only and side-effect free. A singular data source errors on zero or multiple matches; a plural data source (plural noun name) returns zero-or-more as a collection and errors on neither.
• Attached policies/rules, long-running task invocations, and versioned artifacts usually deserve their own resources rather than attributes on the parent.
• Start/stop or enable/disable state belongs as an attribute in the resource, not as a separate resource.

Schema Design

Attribute Types

Give every attribute a MarkdownDescription — tfplugindocs publishes it, and it is the primary user-facing documentation.

Plan Modifiers

// Force replacement when value changes
stringplanmodifier.RequiresReplace()

// Keep a known value during plan instead of (known after apply)
stringplanmodifier.UseStateForUnknown()

Validators

stringvalidator.LengthBetween(1, 255)
stringvalidator.RegexMatches(regexp.MustCompile(`^[a-z0-9-]+$`), "must be lowercase alphanumeric with hyphens")
stringvalidator.OneOf("small", "medium", "large")
int64validator.Between(1, 100)
listvalidator.SizeAtLeast(

Sensitive Attributes

"password": schema.StringAttribute{
    Required:  true,
    Sensitive: true,
},

State Management

Finders

Centralize “get one thing or a typed not-found” in a finder so Read, Delete, waiters, and tests all share identical not-found semantics:

func findWidgetByID(ctx context.Context, client *examplecloud.Client, id string) (*examplecloud.Widget, error) {
    output, err := client.GetWidget(ctx, &examplecloud.

Waiting for Resource States

Many APIs return from Create/Delete before the resource is usable/gone. Use retry.StateChangeConf (from github.com/hashicorp/terraform-plugin-sdk/v2/helper/retry — usable from Framework providers), with a status function built on the finder and timeouts in named constants:

stateConf := &retry.StateChangeConf{
    Pending: []string{"CREATING", "PENDING"},
    Target:  []string{"ACTIVE"},
    Refresh: statusWidget(ctx, r.client, id), // one poll of the finder: (obj, status, err)
    Timeout: widgetCreatedTimeout,
}
outputRaw, err := stateConf.WaitForStateContext(ctx)

The full status/wait function pairs (create and delete waiters, failure-state handling, post-create not-found retries, eventual-consistency patterns) are in references/retries-and-waiters.md — read it whenever the API is asynchronous or eventually consistent.

Testing

Every resource ships with, at minimum:

• _basic — create with minimal config, assert attributes, then an import step (ImportState: true, ImportStateVerify: true)
• _disappears — delete the object out-of-band mid-test; the next plan must propose recreation, not error
• Per-attribute tests — exercise updates for each non-trivial argument

Naming grammar: tests TestAcc{Resource}_{group?}_{description}, helpers testAccCheck{Resource}Exists / testAccCheck{Resource}Destroy, config functions testAcc{Resource}Config_{description}. Keep configs self-contained, randomize real resource names, and never hardcode environment-specific values (account IDs, zones, versions).

func TestAccWidget_basic(t *testing.T) {
    rName := acctest.RandStringFromCharSet(10, acctest.CharSetAlphaNum)
    resourceName := "examplecloud_widget.test"

    resource.ParallelTest(t, resource

Use the provider-test-patterns skill (if available) for the full testing treatment: config helper style (%[1]q indexed verbs), statecheck/plancheck, CompareValue, custom StateCheck implementations for exists/disappears helpers, sweepers, and ephemeral resource testing. Use the run-acceptance-tests skill for executing and debugging test runs.

Error Handling

Match API errors by type, not message text, and wrap with context:

var notFound *examplecloud.NotFoundError
if errors.As(err, &notFound) {
    // resource doesn't exist
}

// Wrapping inside helpers: preserve the cause with %w
return fmt.Errorf("creating Widget (%s): %w", name, err)

Diagnostics follow a consistent grammar — summary names the operation and type, detail carries identifier and cause:

resp.Diagnostics.AddError(
    "Error creating Widget",
    fmt.Sprintf("creating Widget (%s): %s", name, err),
)

resp.Diagnostics.AddAttributeError(
    path.Root("name"),
    "Invalid name",
    "Name must be lowercase alphanumeric",
)

Documentation

Write attribute MarkdownDescriptions first — they are the source of truth. Then generate Registry documentation with tfplugindocs (go generate ./... where wired up), adding docs/**/*.md.tmpl templates only for prose and examples the generator cannot derive. Use the provider-docs skill (if available) for the full documentation workflow and Registry publication rules.

Pre-Submission Checklist

• Not done: Plugin Framework used (no new SDKv2 code)
• Not done: Resource has all CRUD operations implemented
• Not done: Read removes missing resources from state; Delete tolerates already-deleted
• Not done: No redundant id attribute (real API identifier exposed instead)
• Not done: Import implemented and covered by an ImportStateVerify step
• Not done: _basic, _disappears, and per-attribute tests present
• Not done: Waiters used where the API is eventually consistent
• Not done: Error messages name the operation, type, and identifier
• Not done: Sensitive attributes marked; every attribute has a description
• Not done: Docs generated with tfplugindocs
• Not done: Changelog entry added, if the repo tracks release notes (check CONTRIBUTING)

References

• Terraform Plugin Framework (opens in a new tab)
• Resource Development (opens in a new tab)
• Data Source Development (opens in a new tab)
• HashiCorp Provider Design Principles (opens in a new tab)