dao-ai 0.0.27__tar.gz → 0.0.29__tar.gz

dao_ai-0.0.29/.python-version

@@ -0,0 +1 @@
+3.11

{dao_ai-0.0.27 → dao_ai-0.0.29}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dao-ai
-Version: 0.0.27
+Version: 0.0.29
 Summary: DAO AI: A modular, multi-agent orchestration framework for complex AI workflows. Supports agent handoff, tool integration, and dynamic configuration via YAML.
 Project-URL: Homepage, https://github.com/natefleming/dao-ai
 Project-URL: Documentation, https://natefleming.github.io/dao-ai
@@ -18,17 +18,19 @@ Classifier: Intended Audience :: Science/Research
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Topic :: System :: Distributed Computing
-Requires-Python: >=3.12
+Requires-Python: >=3.11
 Requires-Dist: databricks-agents>=1.7.0
 Requires-Dist: databricks-langchain>=0.8.1
 Requires-Dist: databricks-mcp>=0.3.0
 Requires-Dist: databricks-sdk[openai]>=0.67.0
 Requires-Dist: duckduckgo-search>=8.0.2
+Requires-Dist: flashrank>=0.2.8
 Requires-Dist: gepa>=0.0.17
 Requires-Dist: grandalf>=0.8
 Requires-Dist: langchain-mcp-adapters>=0.1.10
@@ -56,7 +58,7 @@ Requires-Dist: tomli>=2.3.0
 Requires-Dist: unitycatalog-ai[databricks]>=0.3.0
 Provides-Extra: databricks
 Requires-Dist: databricks-connect>=15.0.0; extra == 'databricks'
-Requires-Dist: databricks-vectorsearch>=0.56; extra == 'databricks'
+Requires-Dist: databricks-vectorsearch>=0.63; extra == 'databricks'
 Requires-Dist: pyspark>=3.5.0; extra == 'databricks'
 Provides-Extra: dev
 Requires-Dist: mypy>=1.0.0; extra == 'dev'

dao_ai-0.0.29/config/examples/vector_search_with_reranking.yaml

@@ -0,0 +1,293 @@
+# yaml-language-server: $schema=../../schemas/model_config_schema.json
+
+schemas:
+  retail_schema: &retail_schema
+    catalog_name: retail_consumer_goods        # Unity Catalog name
+    schema_name: store_ops                     # Schema within the catalog
+
+resources:
+  llms:
+    # Primary LLM for general tasks
+    default_llm: &default_llm
+      name: databricks-meta-llama-3-3-70b-instruct  # Databricks serving endpoint name
+      temperature: 0.1                              # Low temperature for consistent responses
+      max_tokens: 8192                              # Maximum tokens per response
+
+    embedding_model: &embedding_model
+      name: databricks-gte-large-en                 # Text embedding model
+
+  vector_stores:
+    # Product information vector store for similarity search
+    products_vector_store: &products_vector_store
+      embedding_model: *embedding_model             # Reference to embedding model above
+      endpoint:                                     # Vector search endpoint configuration
+        name: one-env-shared-endpoint-0             # Databricks vector search endpoint
+        type: STANDARD                              # Endpoint type (STANDARD or OPTIMIZED_STORAGE)
+      index:                                        # Vector search index configuration
+        schema: *retail_schema                      # Unity Catalog schema for the index
+        name: retail_vectorstore                    # Index name
+      source_table:                                 # Table containing source data
+        schema: *retail_schema
+        name: products
+      primary_key: item_id                          # Primary key column
+      doc_uri: ~                                    # Optional document URI column (null in this case)
+      embedding_source_column: item_description     # Column to create embeddings from
+      columns:                                      # Columns to include in vector store
+        - item_id
+        - item_name
+        - item_description
+        - category
+        - price
+
+retrievers:
+  # Standard retriever WITHOUT reranking
+  standard_retriever: &standard_retriever
+    vector_store: *products_vector_store            # Reference to vector store defined above
+    columns:                                        # Columns to return in search results
+      - item_id
+      - item_name
+      - item_description
+      - category
+      - price
+    search_parameters:                              # Search configuration
+      num_results: 10                               # Number of results to return
+      filters: {}                                   # Additional filters (empty in this case)
+      query_type: ANN                               # Approximate Nearest Neighbor search
+
+  # Retriever WITH default reranking
+  default_reranked_retriever: &default_reranked_retriever
+    vector_store: *products_vector_store
+    columns:
+      - item_id
+      - item_name
+      - item_description
+      - category
+      - price
+    search_parameters:
+      num_results: 50                               # Retrieve more candidates for reranking
+      filters: {}
+      query_type: ANN
+    rerank: true                                    # Enable reranking with default settings
+
+  # Retriever WITH custom reranking (fast model)
+  fast_reranked_retriever: &fast_reranked_retriever
+    vector_store: *products_vector_store
+    columns:
+      - item_id
+      - item_name
+      - item_description
+      - category
+      - price
+    search_parameters:
+      num_results: 100                              # Cast wide net for reranking
+      filters: {}
+      query_type: ANN
+    rerank:                                         # Custom reranking configuration
+      model: ms-marco-MiniLM-L-6-v2                 # Faster, lighter model
+      top_n: 5                                      # Return top 5 after reranking
+      cache_dir: /tmp/flashrank_cache               # Cache directory for model weights
+
+  # Retriever WITH accurate reranking (hybrid + reranking)
+  accurate_reranked_retriever: &accurate_reranked_retriever
+    vector_store: *products_vector_store
+    columns:
+      - item_id
+      - item_name
+      - item_description
+      - category
+      - price
+    search_parameters:
+      num_results: 50
+      filters: {}
+      query_type: HYBRID                            # Use hybrid search for better recall
+    rerank:
+      model: ms-marco-MiniLM-L-12-v2                # Balanced model (default)
+      top_n: 10                                     # Return top 10 after reranking
+      cache_dir: /tmp/flashrank_cache
+
+tools:
+  # Tool 1: Standard vector search (no reranking)
+  standard_search_tool: &standard_search_tool
+    name: standard_product_search
+    function:
+      type: factory
+      name: dao_ai.tools.create_vector_search_tool
+      args:
+        retriever: *standard_retriever
+        name: standard_product_search
+        description: |
+          Search for products using standard vector similarity search.
+          Returns the 10 most similar products based on embeddings.
+
+  # Tool 2: Vector search with default reranking
+  default_reranked_tool: &default_reranked_tool
+    name: reranked_product_search
+    function:
+      type: factory
+      name: dao_ai.tools.create_vector_search_tool
+      args:
+        retriever: *default_reranked_retriever
+        name: reranked_product_search
+        description: |
+          Search for products with reranking enabled.
+          Retrieves 50 candidates and reranks them using FlashRank
+          with the default model (ms-marco-MiniLM-L-12-v2).
+
+  # Tool 3: Vector search with fast reranking
+  fast_reranked_tool: &fast_reranked_tool
+    name: fast_reranked_search
+    function:
+      type: factory
+      name: dao_ai.tools.create_vector_search_tool
+      args:
+        retriever: *fast_reranked_retriever
+        name: fast_reranked_search
+        description: |
+          Fast product search with lightweight reranking.
+          Retrieves 100 candidates, reranks with a fast model,
+          and returns the top 5 most relevant results.
+
+  # Tool 4: Vector search with accurate reranking
+  accurate_reranked_tool: &accurate_reranked_tool
+    name: accurate_product_search
+    function:
+      type: factory
+      name: dao_ai.tools.create_vector_search_tool
+      args:
+        retriever: *accurate_reranked_retriever
+        name: accurate_product_search
+        description: |
+          High-accuracy product search combining hybrid search and reranking.
+          Uses HYBRID query type for better initial recall, then reranks
+          using a more accurate model to return the best 10 results.
+
+agents:
+  # Agent 1: Uses standard search (no reranking)
+  standard_search_agent: &standard_search_agent
+    name: standard_search
+    description: "Product search agent using standard vector similarity"
+    model: *default_llm
+    tools:
+      - *standard_search_tool
+    prompt: |
+      You are a product search assistant. Use the standard_product_search tool
+      to find products based on customer queries. Provide clear, helpful responses.
+    handoff_prompt: |
+      Handles basic product searches using standard vector similarity.
+
+  # Agent 2: Uses reranking with defaults
+  reranked_search_agent: &reranked_search_agent
+    name: reranked_search
+    description: "Product search agent with reranking for improved accuracy"
+    model: *default_llm
+    tools:
+      - *default_reranked_tool
+    prompt: |
+      You are an enhanced product search assistant with reranking capabilities.
+      Use the reranked_product_search tool to find the most relevant products.
+      The results are reranked for better accuracy.
+    handoff_prompt: |
+      Handles product searches with reranking for improved result quality.
+
+  # Agent 3: Uses fast reranking
+  fast_search_agent: &fast_search_agent
+    name: fast_search
+    description: "Product search agent with fast reranking"
+    model: *default_llm
+    tools:
+      - *fast_reranked_tool
+    prompt: |
+      You are a fast product search assistant. Use the fast_reranked_search tool
+      to quickly find the most relevant products from a large candidate set.
+    handoff_prompt: |
+      Handles product searches with fast reranking for high-volume scenarios.
+
+  # Agent 4: Uses accurate reranking with hybrid search
+  accurate_search_agent: &accurate_search_agent
+    name: accurate_search
+    description: "Product search agent with hybrid search and accurate reranking"
+    model: *default_llm
+    tools:
+      - *accurate_reranked_tool
+    prompt: |
+      You are a high-accuracy product search assistant. Use the accurate_product_search
+      tool which combines hybrid search with reranking for the best possible results.
+    handoff_prompt: |
+      Handles product searches requiring highest accuracy using hybrid search and reranking.
+
+app:
+  name: vector_search_with_reranking_example
+  description: "Demonstrates vector search with and without reranking"
+  log_level: INFO
+  registered_model:
+    schema: *retail_schema
+    name: vector_search_reranking_agent
+  endpoint_name: vector_search_reranking_agent
+  tags:
+    demo: reranking
+    feature: vector_search
+  permissions:
+    - principals: [users]
+      entitlements:
+        - CAN_QUERY
+  agents:
+    # Include all agents to demonstrate different reranking configurations
+    - *standard_search_agent        # No reranking (baseline)
+    - *reranked_search_agent        # Default reranking
+    - *fast_search_agent            # Fast reranking
+    - *accurate_search_agent        # Accurate reranking with hybrid
+  orchestration:
+    supervisor:
+      model: *default_llm
+
+# ==============================================================================
+# RERANKING BEST PRACTICES AND CONFIGURATION GUIDE
+# ==============================================================================
+#
+# This example demonstrates 4 different search configurations:
+#
+# 1. Standard Search (no reranking) - standard_search_agent
+#    - Baseline vector similarity search
+#    - Fast but may return less relevant results
+#    - Use when: Speed is critical, simple queries
+#
+# 2. Default Reranking - reranked_search_agent
+#    - Uses rerank: true for automatic configuration
+#    - Retrieves 50 candidates, reranks with ms-marco-MiniLM-L-12-v2
+#    - Good balance of speed and accuracy
+#    - Use when: Need better results than standard search
+#
+# 3. Fast Reranking - fast_search_agent
+#    - Uses lighter model (ms-marco-MiniLM-L-6-v2)
+#    - Retrieves 100 candidates, returns top 5
+#    - Faster than default but still improves results
+#    - Use when: High volume, need speed with some accuracy boost
+#
+# 4. Accurate Reranking - accurate_search_agent
+#    - Combines HYBRID search with reranking
+#    - Best possible result quality
+#    - Higher latency (~200-500ms)
+#    - Use when: Quality is paramount (customer-facing, critical queries)
+#
+# Configuration Pattern:
+#   retriever:
+#     vector_store: *your_vector_store
+#     search_parameters:
+#       num_results: 50              # Retrieve 2-10x more than needed
+#       query_type: ANN              # or HYBRID for better recall
+#     rerank:                        # Optional reranking
+#       model: ms-marco-MiniLM-L-12-v2  # Choose model
+#       top_n: 10                    # Final count after reranking
+#       cache_dir: /tmp/flashrank_cache
+#
+# Performance Tips:
+#   - First request downloads model (~50MB) - allow extra time
+#   - Subsequent requests are fast (100-500ms)
+#   - More candidates = better reranking but slower
+#   - Cache directory should be on fast storage
+#
+# Model Selection:
+#   - ms-marco-TinyBERT-L-2-v2: Fastest, basic accuracy
+#   - ms-marco-MiniLM-L-6-v2: Fast, good accuracy
+#   - ms-marco-MiniLM-L-12-v2: Balanced (default)
+#   - rank-T5-flan: Most accurate, slowest

{dao_ai-0.0.27 → dao_ai-0.0.29}/config/hardware_store/supervisor.yaml

@@ -108,12 +108,12 @@ resources:
   # Define Unity Catalog tables used by the system
   tables:
     # Product catalog table
-    product_table: &product_table
+    product_table: 
       schema: *retail_schema                        # Reference to schema defined above
       name: products                                # Table name
     
     # Inventory tracking table
-    inventory_table: &inventory_table
+    inventory_table: 
       schema: *retail_schema
       name: inventory
 
@@ -123,7 +123,7 @@ resources:
   # Define Unity Catalog volumes for file storage
   volumes:
     # Volume for retail AI assets and files
-    retail_volume: &retail_volume
+    retail_volume: 
       schema: *retail_schema
       name: retail_ai
 
@@ -162,88 +162,11 @@ resources:
   # Define SQL warehouses for query execution
   warehouses:
     # Shared warehouse for general query execution
-    shared_endpoint_warehouse: &shared_endpoint_warehouse
+    shared_endpoint_warehouse: 
       name: "Shared Endpoint Warehouse"             # Human-readable name
       description: "A warehouse for shared endpoints"  # Description
       warehouse_id: 148ccb90800933a1                # Databricks warehouse ID
 
-  # ---------------------------------------------------------------------------
-  # DATABASES
-  # ---------------------------------------------------------------------------
-  # Configure external databases (e.g., PostgreSQL for checkpoints and memory)
-  databases:
-    # PostgreSQL database for agent memory and checkpoints
-    retail_database: &retail_database
-      name: "retail-consumer-goods"
-      description: "Database for agent memory and checkpoints" 
-      
-      # Database connection parameters
-      # These can be set via environment variables or Databricks secrets
-      host:                                                 
-        default_value: localhost                                             
-        options:
-          - env: PGHOST                              # Environment variable
-          - scope: retail_ai                         # Databricks secret scope
-            secret: PGHOST                           # Secret name
-      port: 
-        default_value: 5432
-        options:
-        - env: PGPORT 
-        - scope: retail_ai
-          secret: PGPORT
-      database: 
-        default_value: databricks_postgres           # Default database name
-        options:
-        - env: PGDATABASE 
-        - scope: retail_ai
-          secret: PGDATABASE
-
-
-      # Traditional PostgreSQL authentication (commented out)
-      # Uncomment and configure if using direct database auth instead of OAuth2
-      # user: 
-      #   default_value: postgres
-      #   options:
-      #   - env: PGUSER 
-      #   - scope: retail_ai
-      #     secret: PGUSER
-      # password: 
-      #   default_value: postgres
-      #   options:
-      #   - env: PGPASSWORD 
-      #   - scope: retail_ai
-      #     secret: PGPASSWORD
-
-      # OAuth2 client credentials for Databricks authentication
-      # Used to generate access tokens for database connections
-      # This approach is recommended for Databricks environments
-      client_id:
-        options:
-          - env: RETAIL_AI_DATABRICKS_CLIENT_ID      # Service principal client ID
-          - scope: retail_ai
-            secret: RETAIL_AI_DATABRICKS_CLIENT_ID
-      client_secret:
-        options:
-          - env: RETAIL_AI_DATABRICKS_CLIENT_SECRET  # Service principal secret
-          - scope: retail_ai
-            secret: RETAIL_AI_DATABRICKS_CLIENT_SECRET
-      workspace_host:
-        options:
-          - env: RETAIL_AI_DATABRICKS_HOST           # Databricks workspace URL
-          - scope: retail_ai
-            secret: RETAIL_AI_DATABRICKS_HOST
-          
-      # Optional: Additional PostgreSQL connection parameters
-      # Uncomment and modify as needed for your specific setup
-      # connection_kwargs:                            # Additional connection parameters
-      #   autocommit: True                            # Enable autocommit mode
-      #   prepare_threshold: 0                        # Disable prepared statements
-      #   sslmode: require                            # Require SSL connection
-      #   connect_timeout: 10                         # Connection timeout in seconds
-      
-      # Optional: Connection pool configuration
-      # max_pool_size: 20                             # Maximum connections in pool
-      # timeout: 5                                    # Connection timeout in seconds
 
 # =============================================================================
 # RETRIEVERS CONFIGURATION
@@ -264,9 +187,15 @@ retrievers:
       - class_cd
       - description
     search_parameters:                              # Search configuration
-      num_results: 10                               # Maximum number of results to return
+      num_results: 50                               # Maximum number of results to return
       filters: {}                                   # Additional filters (empty in this case)
       query_type: ANN                               # Approximate Nearest Neighbor search
+    rerank:
+      top_n: 10
+      columns:
+        - merchandise_class
+        - product_name
+        - brand_name
 
 # =============================================================================
 # TOOLS CONFIGURATION
@@ -310,7 +239,7 @@ tools:
       args: {}                                      # No arguments needed
 
   # Vector search tool for product similarity
-  vector_search_tool: &vector_search_tool
+  find_product_details_by_description_tool: &find_product_details_by_description_tool
     name: vector_search
     function:
       type: factory
@@ -318,17 +247,17 @@ tools:
       args:
         retriever: *products_retriever              # Reference to retriever config
         name: product_vector_search_tool            # Tool instance name
-        description: "Search for products using vector search"  # Tool description
+        description: "Search for products by description"  # Tool description
 
   # Product details search by description
-  find_product_details_by_description_tool:
-    &find_product_details_by_description_tool
-    name: find_product_details_by_description
-    function:
-      type: factory
-      name: hardware_store.tools.find_product_details_by_description_tool
-      args:
-        retriever: *products_retriever              # Uses product retriever for search
+  # find_product_details_by_description_tool:
+  #   &find_product_details_by_description_tool
+  #   name: find_product_details_by_description
+  #   function:
+  #     type: factory
+  #     name: hardware_store.tools.find_product_details_by_description_tool
+  #     args:
+  #       retriever: *products_retriever              # Uses product retriever for search
 
   # ---------------------------------------------------------------------------
   # UNITY CATALOG TOOLS
@@ -361,20 +290,6 @@ tools:
       type: unity_catalog
       <<: *find_inventory_by_upc
 
-  # Store-specific inventory tools using Unity Catalog functions
-  find_store_inventory_by_sku_uc_tool: &find_store_inventory_by_sku_uc_tool
-    name: find_store_inventory_by_sku_uc
-    function:
-      type: unity_catalog
-      <<: *find_store_inventory_by_sku
-
-  find_store_inventory_by_upc_uc_tool: &find_store_inventory_by_upc_uc_tool
-    name: find_store_inventory_by_upc_uc
-    function:
-      type: unity_catalog
-      <<: *find_store_inventory_by_upc
-      
-
       
 # =============================================================================
 # GUARDRAILS CONFIGURATION
@@ -408,26 +323,6 @@ guardrails:
       ### Response:
       {outputs}
 
-# =============================================================================
-# MEMORY CONFIGURATION
-# =============================================================================
-# Configure persistent storage for agent conversations and context
-
-memory: &memory
-  # Conversation checkpointing for state persistence
-  checkpointer: 
-    name: default_checkpointer                      # Checkpointer identifier
-    type: memory                                    # Storage type (postgres or memory)
-    #database: *retail_database                     # Reference to database configuration
-
-  # Long-term memory store for agent context
-  store: 
-    name: default_store                             # Store identifier
-    type: memory                                    # Store type (memory or postgres)
-    embedding_model: *embedding_model               # Model for semantic memory
-    dims: 1536                                      # Embedding dimensions
-   # database: *retail_database                     # Database for persistence (if needed)
-
 # =============================================================================
 # AGENTS CONFIGURATION
 # =============================================================================
@@ -863,32 +758,7 @@ agents:
       Questions asking for product suggestions, recommendations, or "best" products for specific needs
       Example: "What's the best drill for home use?" or "Recommend me a good beginner woodworking kit"
 
-  process_image: &process_image
-    name: process_image
-    description: "Image analysis specialist for extracting product information from images, including summaries, product names, and UPC codes"
-    model: *reasoning_llm
-    tools: []
-    prompt: |
-      ### Task Description
-
-      Analyze the provided product image and extract the following specific information into a structured format:
-
-      - Summary: Provide a concise summary (1-3 sentences) describing what is shown in the image. Focus on describing the product(s), packaging, and any prominent visual elements.
-      - Product Names: Extract the exact product name(s) as they appear on the packaging. Include brand names and specific product variants if visible. List each distinct product separately.
-      - UPC Codes: Identify and extract any UPC (Universal Product Code) barcodes visible in the image. These typically appear as 12-digit numbers beneath a barcode. Provide the complete numeric sequence without dashes or spaces.
-
-      ### Important Guidelines
-
-      - Extract information ONLY if it is clearly visible in the image
-      - For product names, use the EXACT text as it appears on packaging
-      - For UPC codes, verify all digits are clearly legible before extraction
-      - If any requested information is not visible or legible in the image, omit it entirely rather than guessing
-      - Do not provide placeholder values, assumptions, or partial information
-      - Format all extracted information according to the specified output structure
-
-      ### Response Guidelines
-      If any field cannot be determined from the image, exclude it from the output rather than providing an empty value.
-
+ 
 # =============================================================================
 # APPLICATION CONFIGURATION
 # =============================================================================
@@ -919,7 +789,6 @@ app:
     - *recommendation                               # Product recommendation agent
     - *general                                      # General purpose agent
   orchestration:                                    # Agent orchestration configuration
-    #memory: *memory                                 # Reference to memory configuration
     supervisor:                                     # Supervisor orchestration pattern
       model: *tool_calling_llm                      # LLM for routing decisions
   message_hooks: 

{dao_ai-0.0.27 → dao_ai-0.0.29}/environment.yaml

@@ -60,6 +60,7 @@ dependencies:
   - gunicorn==23.0.0
   # Search and tools
   - duckduckgo-search==8.0.4
+  - flashrank==0.2.8
   - litellm==1.75.9
   # Utilities
   - loguru==0.7.3

{dao_ai-0.0.27 → dao_ai-0.0.29}/pyproject.toml

@@ -4,11 +4,11 @@ build-backend = "hatchling.build"
 
 [project]
 name = "dao-ai"
-version = "0.0.27"
+version = "0.0.29"
 description = "DAO AI: A modular, multi-agent orchestration framework for complex AI workflows. Supports agent handoff, tool integration, and dynamic configuration via YAML."
 readme = "README.md"
 license = { text = "MIT" }
-requires-python = ">=3.12"
+requires-python = ">=3.11"
 authors = [
     { name = "Nate Fleming", email = "nate.fleming@databricks.com" },
     { name = "Nate Fleming", email = "nate.fleming@gmail.com" },
@@ -35,6 +35,7 @@ classifiers = [
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent",
     "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
     "Topic :: Scientific/Engineering :: Artificial Intelligence",
@@ -47,6 +48,7 @@ dependencies = [
     "databricks-mcp>=0.3.0",
     "databricks-sdk[openai]>=0.67.0",
     "duckduckgo-search>=8.0.2",
+    "flashrank>=0.2.8",
     "gepa>=0.0.17",
     "grandalf>=0.8",
     "langchain>=0.3.27",
@@ -102,7 +104,7 @@ test = [
     "pytest-mock>=3.10.0",
 ]
 databricks = [
-    "databricks-vectorsearch>=0.56",
+    "databricks-vectorsearch>=0.63",
     "pyspark>=3.5.0",
     "databricks-connect>=15.0.0",
 ]
@@ -165,7 +167,7 @@ exclude = [
 
 [tool.ruff]
 # Enable the Ruff formatter
-target-version = "py312"  # Specify your Python version
+target-version = "py311"  # Minimum supported Python version
 line-length = 88  # Same as Black's default
 indent-width = 4