@dataparade/cli 0.0.2 → 0.0.4

package/patterns/classifier/actors.classifier.yaml

@@ -0,0 +1,17 @@
+---
+# Actor-specific classifier pattern defaults.
+# This file maps actor-related pattern IDs to component type/subType.
+#
+# Loaded alongside classifier.patterns.yaml by src/classifier/config.ts.
+
+pattern_defaults:
+  web_actor:
+    type: actor
+    subType: customer
+    priority: 15
+
+  service_actor:
+    type: actor
+    subType: employee
+    priority: 15
+

package/patterns/classifier/components.classifier.yaml

@@ -0,0 +1,110 @@
+---
+# Configuration for RawFinding → DetectedComponent classification.
+# All mapping values and string lists live here (no hard-coded tables in TS).
+
+name_normalization:
+  # Suffix tokens to strip from component names (case-insensitive).
+  remove_suffixes:
+    - client
+    - service
+    - database
+    - db
+    - api
+    - sdk
+  # Optional regexes applied after basic trimming to remove common suffix patterns.
+  remove_suffix_patterns:
+    - "[-_ ]service$"
+    - "[-_ ]client$"
+    - "[-_ ]api$"
+    - "[-_ ]db$"
+  # Characters to trim from start/end after normalization.
+  trim_chars: " _-"
+
+pattern_defaults:
+  # HTTP routes / controllers → API-like assets.
+  express_route:
+    type: asset
+    subType: api
+    priority: 30
+
+  # Database and cache clients.
+  database_connection:
+    type: asset
+    subType: database
+    priority: 10
+
+  # Generic external HTTP calls when not recognized as known third-party.
+  external_api_call:
+    type: asset
+    subType: api
+    priority: 20
+
+  # Auth middleware / guards (kept as assets for now).
+  auth_middleware:
+    type: asset
+    subType: auth_service
+    priority: 40
+
+  # Configuration access (files/objects).
+  config_file:
+    type: asset
+    subType: config
+    priority: 50
+
+  # Env variable usage.
+  env_variable:
+    type: asset
+    subType: config
+    priority: 60
+
+  # Lambda-style handlers (future use; kept simple for now).
+  lambda_handler:
+    type: asset
+    subType: function
+    priority: 70
+
+  # Container or service definitions.
+  container_service:
+    type: asset
+    subType: container
+    priority: 80
+
+  # Terraform / IaC (static .tf; subtype refined via finding.properties.componentSubType).
+  terraform_resource:
+    type: asset
+    subType: application
+    priority: 14
+
+  terraform_module:
+    type: asset
+    subType: application
+    priority: 16
+
+  terraform_provider:
+    type: third_party
+    subType: cloud_provider
+    priority: 12
+
+database_type_mapping:
+  # Map databaseType hints from TS patterns → component subType.
+  postgres:
+    subType: database
+  mysql:
+    subType: database
+  mongo:
+    subType: database
+  redis:
+    subType: database
+  unknown:
+    subType: database
+
+resolution:
+  # Prefer third_party over asset when both are possible for the same component.
+  prefer_third_party_over_asset: true
+  # How to aggregate confidence across findings in a group: "max" or "average".
+  confidence_aggregation: max
+
+# Env variable keys that should not be emitted as components (runtime/deployment only).
+env_variable_exclude_keys:
+  - NODE_ENV
+

package/patterns/classifier/third-party.classifier.yaml

@@ -0,0 +1,169 @@
+---
+# Known third-party services used by the CLI classifier.
+# DP-P0-CLI-202 – dedicated catalog for third_party components.
+
+third_parties:
+  # Payment processors.
+  - serviceName: stripe
+    matchKeys:
+      - stripe
+    type: third_party
+    subType: payment_processor
+
+  - serviceName: braintree
+    matchKeys:
+      - braintree
+      - "api.braintreepayments.com"
+    type: third_party
+    subType: payment_processor
+
+  - serviceName: adyen
+    matchKeys:
+      - adyen
+      - "checkout.adyen.com"
+    type: third_party
+    subType: payment_processor
+
+  - serviceName: paypal
+    matchKeys:
+      - paypal
+      - "api.paypal.com"
+      - "paypalobjects.com"
+    type: third_party
+    subType: payment_processor
+
+  # Email/SaaS providers.
+  - serviceName: sendgrid
+    matchKeys:
+      - sendgrid
+    type: third_party
+    subType: saas_service
+
+  - serviceName: twilio
+    matchKeys:
+      - twilio
+    type: third_party
+    subType: saas_service
+
+  - serviceName: firebase
+    matchKeys:
+      - firebase
+    type: third_party
+    subType: saas_service
+
+  # Database / BaaS platforms.
+  - serviceName: supabase
+    matchKeys:
+      - supabase
+      - "supabase.co"
+      - "supabase.com"
+    type: third_party
+    subType: saas_service
+
+  # API / AI providers.
+  - serviceName: openai
+    matchKeys:
+      - openai
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: anthropic
+    matchKeys:
+      - anthropic
+      - '@anthropic-ai/'
+      - 'api.anthropic.com'
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: openrouter
+    matchKeys:
+      - openrouter
+      - '@openrouter/'
+      - openrouter.ai
+      - openrouter.com
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: google_ai
+    matchKeys:
+      - google_ai
+      - 'generativelanguage.googleapis.com'
+      - '@google-cloud/vertexai'
+      - vertexai
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: clipdrop
+    matchKeys:
+      - clipdrop
+      - clipdrop-api.co
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: thenextleg
+    matchKeys:
+      - thenextleg
+      - thenextleg.io
+    type: third_party
+    subType: ai_provider
+
+  - serviceName: vectorizer
+    matchKeys:
+      - vectorizer
+      - vectorizer.ai
+    type: third_party
+    subType: ai_provider
+
+  # Cloud providers.
+  - serviceName: aws
+    matchKeys:
+      - aws
+      - "@aws-sdk/"
+    type: third_party
+    subType: cloud_provider
+
+  - serviceName: gcp
+    matchKeys:
+      - gcp
+      - "@google-cloud/"
+      - google-cloud
+    type: third_party
+    subType: cloud_provider
+
+  - serviceName: azure
+    matchKeys:
+      - azure
+      - "@azure/"
+    type: third_party
+    subType: cloud_provider
+
+  # Auth & identity providers.
+  - serviceName: google_auth
+    matchKeys:
+      - "accounts.google.com"
+      - "oauth2.googleapis.com"
+      - "www.googleapis.com/oauth2"
+      - "google-auth-library"
+      - "google_auth_oauthlib"
+      - "@react-oauth/google"
+      - "googleapis"
+    type: third_party
+    subType: saas_service
+
+  - serviceName: auth0
+    matchKeys:
+      - auth0
+      - "auth0.com"
+      - "AUTH0_"
+    type: third_party
+    subType: saas_service
+
+  # Error monitoring / observability.
+  - serviceName: sentry
+    matchKeys:
+      - sentry
+      - "sentry.io"
+      - "sentry_sdk"
+    type: third_party
+    subType: saas_service
+

package/patterns/kubernetes-terraform-catalog.snapshot.json

@@ -0,0 +1,94 @@
+{
+  "generatedAt": "2026-05-20T15:18:04.013Z",
+  "source": "@cdktf/provider-kubernetes@11.x (HashiCorp kubernetes)",
+  "resourceTypes": [
+    "kubernetes_all_namespaces",
+    "kubernetes_annotations",
+    "kubernetes_api_service",
+    "kubernetes_api_service_v1",
+    "kubernetes_certificate_signing_request",
+    "kubernetes_certificate_signing_request_v1",
+    "kubernetes_cluster_role",
+    "kubernetes_cluster_role_binding",
+    "kubernetes_cluster_role_binding_v1",
+    "kubernetes_cluster_role_v1",
+    "kubernetes_config_map",
+    "kubernetes_config_map_v1",
+    "kubernetes_config_map_v1_data",
+    "kubernetes_cron_job",
+    "kubernetes_cron_job_v1",
+    "kubernetes_csi_driver",
+    "kubernetes_csi_driver_v1",
+    "kubernetes_daemon_set_v1",
+    "kubernetes_daemonset",
+    "kubernetes_default_service_account",
+    "kubernetes_default_service_account_v1",
+    "kubernetes_deployment",
+    "kubernetes_deployment_v1",
+    "kubernetes_endpoint_slice_v1",
+    "kubernetes_endpoints",
+    "kubernetes_endpoints_v1",
+    "kubernetes_env",
+    "kubernetes_horizontal_pod_autoscaler",
+    "kubernetes_horizontal_pod_autoscaler_v1",
+    "kubernetes_horizontal_pod_autoscaler_v2",
+    "kubernetes_horizontal_pod_autoscaler_v2beta2",
+    "kubernetes_ingress",
+    "kubernetes_ingress_class",
+    "kubernetes_ingress_class_v1",
+    "kubernetes_ingress_v1",
+    "kubernetes_job",
+    "kubernetes_job_v1",
+    "kubernetes_labels",
+    "kubernetes_limit_range",
+    "kubernetes_limit_range_v1",
+    "kubernetes_manifest",
+    "kubernetes_mutating_webhook_configuration",
+    "kubernetes_mutating_webhook_configuration_v1",
+    "kubernetes_namespace",
+    "kubernetes_namespace_v1",
+    "kubernetes_network_policy",
+    "kubernetes_network_policy_v1",
+    "kubernetes_node_taint",
+    "kubernetes_nodes",
+    "kubernetes_persistent_volume",
+    "kubernetes_persistent_volume_claim",
+    "kubernetes_persistent_volume_claim_v1",
+    "kubernetes_persistent_volume_v1",
+    "kubernetes_pod",
+    "kubernetes_pod_disruption_budget",
+    "kubernetes_pod_disruption_budget_v1",
+    "kubernetes_pod_security_policy",
+    "kubernetes_pod_security_policy_v1beta1",
+    "kubernetes_pod_v1",
+    "kubernetes_priority_class",
+    "kubernetes_priority_class_v1",
+    "kubernetes_provider",
+    "kubernetes_replication_controller",
+    "kubernetes_replication_controller_v1",
+    "kubernetes_resource",
+    "kubernetes_resource_quota",
+    "kubernetes_resource_quota_v1",
+    "kubernetes_resources",
+    "kubernetes_role",
+    "kubernetes_role_binding",
+    "kubernetes_role_binding_v1",
+    "kubernetes_role_v1",
+    "kubernetes_runtime_class_v1",
+    "kubernetes_secret",
+    "kubernetes_secret_v1",
+    "kubernetes_secret_v1_data",
+    "kubernetes_server_version",
+    "kubernetes_service",
+    "kubernetes_service_account",
+    "kubernetes_service_account_v1",
+    "kubernetes_service_v1",
+    "kubernetes_stateful_set",
+    "kubernetes_stateful_set_v1",
+    "kubernetes_storage_class",
+    "kubernetes_storage_class_v1",
+    "kubernetes_token_request_v1",
+    "kubernetes_validating_webhook_configuration",
+    "kubernetes_validating_webhook_configuration_v1"
+  ]
+}

package/patterns/kubernetes-terraform-service-hints.generated.json

@@ -0,0 +1,258 @@
+{
+  "generatedAt": "2026-05-20T15:18:04.013Z",
+  "source": "@cdktf/provider-kubernetes@11.x (HashiCorp kubernetes)",
+  "hints": [
+    {
+      "id": "gen_kubernetes_all_family",
+      "resource_type_regex": "^kubernetes_all_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_annotations_family",
+      "resource_type_regex": "^kubernetes_annotations_",
+      "componentSubType": "config",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_api_family",
+      "resource_type_regex": "^kubernetes_api_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_certificate_family",
+      "resource_type_regex": "^kubernetes_certificate_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_cluster_family",
+      "resource_type_regex": "^kubernetes_cluster_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_config_family",
+      "resource_type_regex": "^kubernetes_config_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_cron_family",
+      "resource_type_regex": "^kubernetes_cron_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_csi_family",
+      "resource_type_regex": "^kubernetes_csi_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_daemon_family",
+      "resource_type_regex": "^kubernetes_daemon_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_daemonset_family",
+      "resource_type_regex": "^kubernetes_daemonset_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_default_family",
+      "resource_type_regex": "^kubernetes_default_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_deployment_family",
+      "resource_type_regex": "^kubernetes_deployment_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_endpoint_family",
+      "resource_type_regex": "^kubernetes_endpoint_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_endpoints_family",
+      "resource_type_regex": "^kubernetes_endpoints_",
+      "componentSubType": "api",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_env_family",
+      "resource_type_regex": "^kubernetes_env_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_horizontal_family",
+      "resource_type_regex": "^kubernetes_horizontal_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_ingress_family",
+      "resource_type_regex": "^kubernetes_ingress_",
+      "componentSubType": "api",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_job_family",
+      "resource_type_regex": "^kubernetes_job_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_labels_family",
+      "resource_type_regex": "^kubernetes_labels_",
+      "componentSubType": "config",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_limit_family",
+      "resource_type_regex": "^kubernetes_limit_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_manifest_family",
+      "resource_type_regex": "^kubernetes_manifest_",
+      "componentSubType": "application",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_mutating_family",
+      "resource_type_regex": "^kubernetes_mutating_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_namespace_family",
+      "resource_type_regex": "^kubernetes_namespace_",
+      "componentSubType": "container",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_network_family",
+      "resource_type_regex": "^kubernetes_network_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_node_family",
+      "resource_type_regex": "^kubernetes_node_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_nodes_family",
+      "resource_type_regex": "^kubernetes_nodes_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_persistent_family",
+      "resource_type_regex": "^kubernetes_persistent_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_pod_family",
+      "resource_type_regex": "^kubernetes_pod_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_priority_family",
+      "resource_type_regex": "^kubernetes_priority_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_provider_family",
+      "resource_type_regex": "^kubernetes_provider_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_replication_family",
+      "resource_type_regex": "^kubernetes_replication_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_resource_family",
+      "resource_type_regex": "^kubernetes_resource_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_resources_family",
+      "resource_type_regex": "^kubernetes_resources_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_role_family",
+      "resource_type_regex": "^kubernetes_role_",
+      "componentSubType": "auth_service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_runtime_family",
+      "resource_type_regex": "^kubernetes_runtime_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_secret_family",
+      "resource_type_regex": "^kubernetes_secret_",
+      "componentSubType": "config",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_server_family",
+      "resource_type_regex": "^kubernetes_server_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_service_family",
+      "resource_type_regex": "^kubernetes_service_",
+      "componentSubType": "api",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_stateful_family",
+      "resource_type_regex": "^kubernetes_stateful_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_storage_family",
+      "resource_type_regex": "^kubernetes_storage_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_token_family",
+      "resource_type_regex": "^kubernetes_token_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    },
+    {
+      "id": "gen_kubernetes_validating_family",
+      "resource_type_regex": "^kubernetes_validating_",
+      "componentSubType": "service",
+      "cloud_provider": "kubernetes"
+    }
+  ]
+}