> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/microsoft/graphrag/llms.txt
> Use this file to discover all available pages before exploring further.

# Settings reference

> Complete reference for GraphRAG settings.yaml configuration

GraphRAG can be configured using a `settings.yaml` or `settings.json` file in your project root. This page documents all available configuration options.

## Environment variable substitution

Configuration values can reference environment variables using `${VAR_NAME}` syntax:

```yaml theme={null}
completion_models:
  default_completion_model:
    api_key: ${GRAPHRAG_API_KEY}
```

```bash .env theme={null}
GRAPHRAG_API_KEY=your-api-key-here
```

<Note>
  If a `.env` file is present in your project root, environment variables will be automatically loaded.
</Note>

## Language model configuration

### Completion models

Define completion models for text generation tasks:

```yaml theme={null}
completion_models:
  default_completion_model:
    model_provider: openai
    model: gpt-4.1
    auth_method: api_key
    api_key: ${GRAPHRAG_API_KEY}
    retry:
      type: exponential_backoff
      max_retries: 7
      base_delay: 2.0
    rate_limit:
      type: sliding_window
      period_in_seconds: 60
      requests_per_period: 100
```

<Tip>
  You can define multiple models and reference them by key in workflow configurations. See [LLM Models](/configuration/llm-models) for detailed examples.
</Tip>

### Embedding models

Define embedding models for vector generation:

```yaml theme={null}
embedding_models:
  default_embedding_model:
    model_provider: openai
    model: text-embedding-3-large
    auth_method: api_key
    api_key: ${GRAPHRAG_API_KEY}
```

## Input configuration

### Input settings

Configure document input format and location:

```yaml theme={null}
input:
  type: text  # text, csv, json, jsonl
  encoding: utf-8
  file_pattern: ".*\\.txt$"  # regex pattern for files
  id_column: id  # for CSV/JSON
  title_column: title  # for CSV/JSON
  text_column: text  # for CSV/JSON
```

<ParamField path="type" type="string" default="text">
  Input data format: `text`, `csv`, `json`, or `jsonl`
</ParamField>

<ParamField path="encoding" type="string" default="utf-8">
  Character encoding for input files
</ParamField>

<ParamField path="file_pattern" type="string">
  Regex pattern to match input files (defaults based on type)
</ParamField>

<ParamField path="id_column" type="string">
  Column name for document IDs (CSV/JSON only)
</ParamField>

<ParamField path="title_column" type="string">
  Column name for document titles (CSV/JSON only)
</ParamField>

<ParamField path="text_column" type="string">
  Column name for document text content (CSV/JSON only)
</ParamField>

### Chunking configuration

Configure how documents are split into chunks:

```yaml theme={null}
chunking:
  type: tokens  # tokens or sentence
  size: 1200
  overlap: 100
  encoding_model: o200k_base
  prepend_metadata:
    - title
    - source
```

<ParamField path="type" type="string" default="tokens">
  Chunking strategy: `tokens` or `sentence`
</ParamField>

<ParamField path="size" type="integer" default="1200">
  Maximum chunk size in tokens
</ParamField>

<ParamField path="overlap" type="integer" default="100">
  Number of overlapping tokens between chunks
</ParamField>

<ParamField path="encoding_model" type="string" default="o200k_base">
  Tokenizer model for splitting text
</ParamField>

<ParamField path="prepend_metadata" type="array">
  Document metadata fields to prepend to each chunk
</ParamField>

## Storage configuration

See the [Storage](/configuration/storage) page for detailed storage configuration.

## Workflow configurations

### Text embedding

Configure text embedding generation:

```yaml theme={null}
embed_text:
  embedding_model_id: default_embedding_model
  model_instance_name: text_embedding
  batch_size: 16
  batch_max_tokens: 8191
  names:
    - text_unit_text
    - entity_description
    - community_full_content
```

<ParamField path="embedding_model_id" type="string" required>
  Reference to embedding model configuration
</ParamField>

<ParamField path="batch_size" type="integer" default="16">
  Maximum number of texts to embed in one batch
</ParamField>

<ParamField path="batch_max_tokens" type="integer" default="8191">
  Maximum total tokens per batch
</ParamField>

<ParamField path="names" type="array">
  Which embeddings to generate: `text_unit_text`, `entity_description`, `community_full_content`
</ParamField>

### Graph extraction

Configure LLM-based entity and relationship extraction:

```yaml theme={null}
extract_graph:
  completion_model_id: default_completion_model
  model_instance_name: extract_graph
  prompt: "prompts/extract_graph.txt"
  entity_types:
    - organization
    - person
    - geo
    - event
  max_gleanings: 1
```

<ParamField path="completion_model_id" type="string" required>
  Reference to completion model configuration
</ParamField>

<ParamField path="prompt" type="string">
  Path to extraction prompt file
</ParamField>

<ParamField path="entity_types" type="array" default="[organization, person, geo, event]">
  List of entity types to extract
</ParamField>

<ParamField path="max_gleanings" type="integer" default="1">
  Number of additional extraction passes for thoroughness
</ParamField>

### NLP-based extraction

Configure NLP-based graph extraction (alternative to LLM):

```yaml theme={null}
extract_graph_nlp:
  normalize_edge_weights: true
  concurrent_requests: 25
  async_mode: threaded  # threaded or asyncio
  text_analyzer:
    extractor_type: regex_english  # regex_english, syntactic_parser, cfg
    max_word_length: 15
    include_named_entities: true
```

### Description summarization

Configure entity and relationship description summarization:

```yaml theme={null}
summarize_descriptions:
  completion_model_id: default_completion_model
  model_instance_name: summarize_descriptions
  prompt: "prompts/summarize_descriptions.txt"
  max_length: 500
  max_input_tokens: 4000
```

<ParamField path="max_length" type="integer" default="500">
  Maximum output tokens per summary
</ParamField>

<ParamField path="max_input_tokens" type="integer" default="4000">
  Maximum input tokens to collect for summarization
</ParamField>

### Graph clustering

Configure Leiden hierarchical clustering:

```yaml theme={null}
cluster_graph:
  max_cluster_size: 10
  use_lcc: true
  seed: 0xDEADBEEF
```

<ParamField path="max_cluster_size" type="integer" default="10">
  Maximum cluster size for export
</ParamField>

<ParamField path="use_lcc" type="boolean" default="true">
  Whether to use only the largest connected component
</ParamField>

<ParamField path="seed" type="integer" default="0xDEADBEEF">
  Random seed for consistent clustering results
</ParamField>

### Graph pruning

Configure optional graph pruning to optimize modularity:

```yaml theme={null}
prune_graph:
  min_node_freq: 2
  max_node_freq_std: null
  min_node_degree: 1
  max_node_degree_std: null
  min_edge_weight_pct: 40.0
  remove_ego_nodes: true
  lcc_only: false
```

<ParamField path="min_node_freq" type="integer" default="2">
  Minimum node frequency to retain
</ParamField>

<ParamField path="min_node_degree" type="integer" default="1">
  Minimum node degree (connections) to retain
</ParamField>

<ParamField path="min_edge_weight_pct" type="float" default="40.0">
  Minimum edge weight percentile to retain
</ParamField>

<ParamField path="remove_ego_nodes" type="boolean" default="true">
  Remove ego nodes (nodes connected to everything)
</ParamField>

### Community reports

Configure community report generation:

```yaml theme={null}
community_reports:
  completion_model_id: default_completion_model
  model_instance_name: community_reporting
  graph_prompt: "prompts/community_report_graph.txt"
  text_prompt: "prompts/community_report_text.txt"
  max_length: 2000
  max_input_length: 8000
```

<ParamField path="graph_prompt" type="string">
  Prompt for graph-based community summarization
</ParamField>

<ParamField path="text_prompt" type="string">
  Prompt for text-based community summarization
</ParamField>

<ParamField path="max_length" type="integer" default="2000">
  Maximum output tokens per report
</ParamField>

<ParamField path="max_input_length" type="integer" default="8000">
  Maximum input tokens for report generation
</ParamField>

### Claim extraction

Configure optional claim extraction:

```yaml theme={null}
extract_claims:
  enabled: false
  completion_model_id: default_completion_model
  prompt: "prompts/extract_claims.txt"
  description: "Any claims or facts that could be relevant to information discovery."
  max_gleanings: 1
```

<Warning>
  Claim extraction is disabled by default and requires prompt tuning for your specific domain.
</Warning>

### Snapshots

Configure optional data snapshots:

```yaml theme={null}
snapshots:
  embeddings: false
  graphml: false
  raw_graph: false
```

<ParamField path="embeddings" type="boolean" default="false">
  Export embeddings to parquet files
</ParamField>

<ParamField path="graphml" type="boolean" default="false">
  Export graph to GraphML format
</ParamField>

<ParamField path="raw_graph" type="boolean" default="false">
  Export raw extracted graph before merging
</ParamField>

## Query configurations

### Local search

Configure local search for targeted queries:

```yaml theme={null}
local_search:
  completion_model_id: default_completion_model
  embedding_model_id: default_embedding_model
  prompt: "prompts/local_search_system_prompt.txt"
  text_unit_prop: 0.5
  community_prop: 0.15
  conversation_history_max_turns: 5
  top_k_entities: 10
  top_k_relationships: 10
  max_context_tokens: 12000
```

### Global search

Configure global search for broad queries:

```yaml theme={null}
global_search:
  completion_model_id: default_completion_model
  map_prompt: "prompts/global_search_map_system_prompt.txt"
  reduce_prompt: "prompts/global_search_reduce_system_prompt.txt"
  knowledge_prompt: "prompts/global_search_knowledge_system_prompt.txt"
  data_max_tokens: 12000
  map_max_length: 1000
  reduce_max_length: 2000
  dynamic_search_threshold: 1
```

### DRIFT search

Configure DRIFT search for iterative exploration:

```yaml theme={null}
drift_search:
  completion_model_id: default_completion_model
  embedding_model_id: default_embedding_model
  prompt: "prompts/drift_search_system_prompt.txt"
  reduce_prompt: "prompts/drift_search_reduce_prompt.txt"
  n_depth: 3
  drift_k_followups: 20
  concurrency: 32
```

### Basic search

Configure basic vector search:

```yaml theme={null}
basic_search:
  completion_model_id: default_completion_model
  embedding_model_id: default_embedding_model
  prompt: "prompts/basic_search_system_prompt.txt"
  k: 10
  max_context_tokens: 12000
```

## Advanced settings

### Workflows

Override the default workflow execution order:

```yaml theme={null}
workflows:
  - extract_graph
  - summarize_descriptions
  - cluster_graph
  - community_reports
```

<Info>
  Most users don't need to customize workflows. Only specify this if you want precise control over execution order.
</Info>

### Concurrency settings

Control global concurrency for async operations:

```yaml theme={null}
concurrent_requests: 25
async_mode: threaded  # threaded or asyncio
```

## Next steps

<CardGroup cols={2}>
  <Card title="LLM models" icon="brain" href="/configuration/llm-models">
    Detailed guide to configuring language models
  </Card>

  <Card title="Storage" icon="database" href="/configuration/storage">
    Configure storage backends and caching
  </Card>

  <Card title="Initialization" icon="rocket" href="/configuration/initialization">
    Learn about the init command
  </Card>

  <Card title="Start indexing" icon="play" href="/indexing/overview">
    Begin processing your documents
  </Card>
</CardGroup>