chore: registry hygiene pass — clear all 34 shadow warnings, fix 33 example.yaml validation errors

L1 validation harness scorecard:
  PASS:   487  →  562  (+75)
  WARN:    33  →    0  (−33)
  FAIL:    74  →   41  (−33)

Changes — shadow-attr warnings (34 components):
  All components with `model:` or `schema:` fields shadowing the
  Pydantic / Dagster Resolvable / Model parent class attributes were
  renamed to `model_id` / `schema_name` with `Field(alias="<orig>")`
  + `model_config = ConfigDict(populate_by_name=True)`. YAML-facing
  field name unchanged — existing example.yaml + customer defs.yaml
  keep working without any rewrite.

  Pattern applied (33 model rename + 4 schema rename) via the new
  tools/fix_shadow_warnings.py script, which reads the L1 validation
  report and applies the transform with sanity-checks (alias presence,
  ConfigDict import, body multiline-preserving rewrite).

Changes — example.yaml validation errors (33 components fixed of 74):
  - 9 components had `source_asset:` in example.yaml but the component
    expected `upstream_asset_key:`. Renamed in-place.
  - 26 components were entirely missing `upstream_asset_key:` from
    example.yaml; added with descriptive per-component placeholders
    (raw_documents, raw_text, raw_tickets, rows_to_index, etc.).
  - 4 components used `*_asset` field names (google_ads_asset,
    facebook_ads_asset, campaign_data_asset, etc.) instead of the
    component's actual `*_asset_key` field. Renamed.
  - 4 IO managers (postgres / databricks / redshift / trino) used
    `schema_name:` in example.yaml but the component field is
    `default_schema:`. Renamed.
  - 5 source / sink components had skeleton example.yamls
    ("See schema.json for full attribute reference") — fleshed out
    with all required fields (cloudwatch_logs_insights_query,
    cloudwatch_metrics_query, dataframe_to_otlp_logs / _metrics / _traces).
  - 3 example.yamls had a misplaced `database_url_env_var:` line from
    a stub template — removed.

Remaining 41 FAILs are diverse one-off issues (missing endpoint /
client_id / api_token_env_var fields; deps wrong-type; renamed
fields like source_table → table_name). These need per-component
attention and will be addressed in a follow-up pass.

New tool:
  tools/fix_shadow_warnings.py — batch-applies the alias-rename
  pattern from the L1 report. Reusable for future shadow warnings
  introduced by Pydantic / Dagster Resolvable API additions.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: eric-thomas-dagster

assets/ai/anthropic_llm/component.py

@@ -25,7 +25,7 @@
     Output,
     MetadataValue,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -169,6 +169,7 @@ class AnthropicLLMComponent(Component, Model, Resolvable):
         ```
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(
         description="Name of the asset that will hold the enriched data"
     )
@@ -177,7 +178,8 @@ class AnthropicLLMComponent(Component, Model, Resolvable):
         description="Anthropic API key. Use ${ANTHROPIC_API_KEY} for environment variables."
     )
 
-    model: str = Field(
+    model_id: str = Field(
+        alias="model",
         default="claude-3-5-sonnet-20241022",
         description="Claude model: claude-3-5-sonnet-20241022, claude-3-opus-20240229, claude-3-sonnet-20240229, claude-3-haiku-20240307"
     )
@@ -393,7 +395,7 @@ class AnthropicLLMComponent(Component, Model, Resolvable):
     def build_defs(self, context: ComponentLoadContext) -> Definitions:
         asset_name = self.asset_name
         api_key = self.api_key
-        model = self.model
+        model = self.model_id
         system_prompt = self.system_prompt
         user_prompt_template = self.user_prompt_template
         input_column = self.input_column

assets/ai/anthropic_llm/example.yaml

@@ -1,6 +1,7 @@
 type: dagster_component_templates.AnthropicLLMComponent
 attributes:
   asset_name: document_analysis
+  upstream_asset_key: raw_documents
 
   # Anthropic Configuration
   api_key: "${ANTHROPIC_API_KEY}"

assets/ai/bank_statement_extractor/component.py

@@ -18,7 +18,7 @@
     Resolvable,
     asset,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -143,6 +143,7 @@ class BankStatementExtractorComponent(Component, Model, Resolvable):
     - Lending underwriting pipelines
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(description="Output Dagster asset name")
     upstream_asset_key: str = Field(
         description="Upstream asset key providing a DataFrame with bank statement content"
@@ -155,7 +156,9 @@ class BankStatementExtractorComponent(Component, Model, Resolvable):
         default="text",
         description="Input type: 'text' (raw text) or 'file' (read file content)",
     )
-    model: str = Field(default="gpt-4o-mini", description="LiteLLM model string")
+    model_id: str = Field(
+        alias="model",
+        default="gpt-4o-mini", description="LiteLLM model string")
     api_key_env_var: str = Field(
         default="OPENAI_API_KEY",
         description="Env var name for API key",
@@ -299,7 +302,7 @@ def build_defs(self, load_context: ComponentLoadContext) -> Definitions:
         upstream_asset_key = self.upstream_asset_key
         input_column = self.input_column
         input_type = self.input_type
-        model = self.model
+        model = self.model_id
         api_key_env_var = self.api_key_env_var
         output_fields = self.output_fields
         batch_size = self.batch_size

assets/ai/contract_extractor/component.py

@@ -18,7 +18,7 @@
     Resolvable,
     asset,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -143,6 +143,7 @@ class ContractExtractorComponent(Component, Model, Resolvable):
     - Vendor agreement analysis
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(description="Output Dagster asset name")
     upstream_asset_key: str = Field(
         description="Upstream asset key providing a DataFrame with contract content"
@@ -155,7 +156,9 @@ class ContractExtractorComponent(Component, Model, Resolvable):
         default="text",
         description="Input type: 'text' (raw text) or 'file' (read file content)",
     )
-    model: str = Field(default="gpt-4o-mini", description="LiteLLM model string")
+    model_id: str = Field(
+        alias="model",
+        default="gpt-4o-mini", description="LiteLLM model string")
     api_key_env_var: str = Field(
         default="OPENAI_API_KEY",
         description="Env var name for API key",
@@ -299,7 +302,7 @@ def build_defs(self, load_context: ComponentLoadContext) -> Definitions:
         upstream_asset_key = self.upstream_asset_key
         input_column = self.input_column
         input_type = self.input_type
-        model = self.model
+        model = self.model_id
         api_key_env_var = self.api_key_env_var
         output_fields = self.output_fields
         batch_size = self.batch_size

assets/ai/conversation_memory/example.yaml

@@ -12,6 +12,7 @@
 type: dagster_component_templates.ConversationMemoryComponent
 attributes:
   asset_name: conversation
+  upstream_asset_key: raw_messages
   memory_file: ./conversation.json
   max_messages: 10
   group_name: conversations

assets/ai/document_chunker/example.yaml

@@ -1,6 +1,7 @@
 type: dagster_component_templates.DocumentChunkerComponent
 attributes:
   asset_name: document_chunks
+  upstream_asset_key: raw_documents
 
   # Chunking Strategy
   strategy: "recursive"  # fixed, recursive, sentence, token_aware, semantic

assets/ai/document_summarizer/component.py

@@ -20,7 +20,7 @@
     Model,
     MetadataValue,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -145,12 +145,15 @@ class DocumentSummarizerComponent(Component, Model, Resolvable):
         ```
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(description="Name of the asset")
     upstream_asset_key: str = Field(description="Upstream asset key providing a DataFrame with documents to summarize")
     input_column: str = Field(default="text", description="Column name containing document text to summarize")
     output_column: str = Field(default="summary", description="Column name for generated summaries")
     provider: str = Field(description="LLM provider")
-    model: str = Field(description="Model name")
+    model_id: str = Field(
+        alias="model",
+        description="Model name")
     summary_type: str = Field(default="concise", description="Type: 'concise', 'detailed', 'bullet_points', 'executive'")
     max_length: Optional[int] = Field(default=None, description="Max summary length in words")
     chunk_size: int = Field(default=3000, description="Chunk size for long documents")
@@ -275,7 +278,7 @@ def build_defs(self, context: ComponentLoadContext) -> Definitions:
         input_column = self.input_column
         output_column = self.output_column
         provider = self.provider
-        model = self.model
+        model = self.model_id
         summary_type = self.summary_type
         max_length = self.max_length
         chunk_size = self.chunk_size

assets/ai/document_summarizer/example.yaml

@@ -4,6 +4,7 @@
 type: dagster_component_templates.DocumentSummarizerComponent
 attributes:
   asset_name: report_summary
+  upstream_asset_key: raw_documents
   provider: openai
   model: gpt-4
   summary_type: executive

assets/ai/dspy_program/component.py

@@ -17,7 +17,7 @@
     Resolvable,
     asset,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -142,13 +142,16 @@ class DspyProgramComponent(Component, Model, Resolvable):
         ```
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(description="Output Dagster asset name")
     upstream_asset_key: str = Field(description="Upstream asset key providing a DataFrame")
     text_column: str = Field(description="Column containing input text")
     output_column: str = Field(default="dspy_output", description="Primary output column (first output field)")
     program_type: str = Field(default="chain_of_thought", description="DSPy program type: predict, chain_of_thought, react, or multi_chain_comparison")
     signature: str = Field(description='DSPy signature string e.g. "question -> answer" or "document -> summary, key_points"')
-    model: str = Field(default="gpt-4o-mini", description="LLM model string")
+    model_id: str = Field(
+        alias="model",
+        default="gpt-4o-mini", description="LLM model string")
     api_key_env_var: Optional[str] = Field(default=None, description="Env var name for API key")
     group_name: Optional[str] = Field(default=None, description="Dagster asset group name")
     partition_type: Optional[str] = Field(
@@ -276,7 +279,7 @@ def build_defs(self, load_context: ComponentLoadContext) -> Definitions:
         output_column = self.output_column
         program_type = self.program_type
         signature = self.signature
-        model = self.model
+        model = self.model_id
         api_key_env_var = self.api_key_env_var
         group_name = self.group_name
 

assets/ai/embeddings_generator/component.py

@@ -22,7 +22,7 @@
     Output,
     MetadataValue,
 )
-from pydantic import Field
+from pydantic import ConfigDict, Field
 
 
 def _build_partitions_def(
@@ -163,6 +163,7 @@ class EmbeddingsGeneratorComponent(Component, Model, Resolvable):
         ```
     """
 
+    model_config = ConfigDict(populate_by_name=True)
     asset_name: str = Field(
         description="Name of the asset that will hold the embeddings"
     )
@@ -171,7 +172,8 @@ class EmbeddingsGeneratorComponent(Component, Model, Resolvable):
         description="Embedding provider: openai, cohere, sentence_transformers, huggingface"
     )
 
-    model: str = Field(
+    model_id: str = Field(
+        alias="model",
         description="Model name (e.g., 'text-embedding-3-small', 'embed-english-v3.0', 'all-MiniLM-L6-v2')"
     )
 
@@ -361,7 +363,7 @@ class EmbeddingsGeneratorComponent(Component, Model, Resolvable):
     def build_defs(self, context: ComponentLoadContext) -> Definitions:
         asset_name = self.asset_name
         provider = self.provider
-        model = self.model
+        model = self.model_id
         api_key = self.api_key
         input_column = self.input_column
         output_column = self.output_column