- Graphwise Platform Documentation
- GraphRAG
- GraphRAG
- The Components Service API
- The Components Service API Reference
The Components Service API Reference
09/02/2026
This topic gives you an overview of endpoints used by GraphRAG along with sample requests and responses as well as query parameters.
API Reference
Method | Endpoint | Details |
|---|---|---|
POST |
| Semantic similarity search |
POST |
| Generate text embedding |
GET |
| Detailed health status |
GET |
| Quick readiness check |
GET |
| Service information |
GET |
| Build information |
Performs semantic similarity search against the vector store.
POST /components/vector/search
Parameter | Type | Required | Default | Details |
|---|---|---|---|---|
text | string | Yes | — | The search query |
top_k | integer | No | 5 | Number of results to return (minimum: 1) |
index | string | No | *configured* | Override target index |
curl -X POST "https://api.example.com/components/vector/search?text=knowledge%20graph&top_k=3"
[
{
"score": 0.8234,
"metadata": {
"AMAZON_BEDROCK_TEXT": "A knowledge graph represents information
as interconnected entities and relationships...",
"x-amz-bedrock-kb-source-uri": "s3://my-bucket/docs/knowledge-graphs.txt",
"x-amz-bedrock-kb-data-source-id": "XIY1CD9O3U",
"id": "chunk-abc123"
},
"id": "1%3A0%3Aabc123"
},
{
"score": 0.7456,
"metadata": {
"AMAZON_BEDROCK_TEXT": "Graph databases store data as nodes and edges,
enabling complex relationship queries...",
"x-amz-bedrock-kb-source-uri": "s3://my-bucket/docs/graph-databases.txt",
"x-amz-bedrock-kb-data-source-id": "XIY1CD9O3U",
"id": "chunk-def456"
},
"id": "1%3A0%3Adef456"
}
]Field | Type | Details |
|---|---|---|
score | float | Similarity score (0-1, higher is better) |
metadata | object | Document metadata |
metadata.AMAZON_BEDROCK_TEXT | string | The text content of this chunk |
metadata.x-amz-bedrock-kb-source-uri | string | Source document S3 URI |
metadata.id | string | Unique chunk identifier |
id | string | OpenSearch document ID |
HTTP Code | Error | Cause |
|---|---|---|
400 | Bad Request | the server could not understand the request because of invalid syntax; missing text parameter |
500 | Internal Server Error | AWS/OpenSearch connection failed |
Generates a vector embedding for the provided text.
POST /components/vector/embed
Parameter | Type | Required | Details |
|---|---|---|---|
text | string | Yes | Text to be converted to an embedding |
curl -X POST "https://api.example.com/components/vector/embed?text=GraphRAG%20enhances%20LLM%20responses"
Returns an array of floating-point numbers (dimensions depend on the configured model).
[-0.011604309, -0.020996094, -0.01058197, -0.007511139, -0.023483276, ...]
The Components Service supports two embedding providers:
Provider | Environment Variable | Details |
|---|---|---|
AWS | EMBEDDING_PROVIDER=aws | AWS Bedrock embedding models (default) |
OpenAI | EMBEDDING_PROVIDER=openai | OpenAI embedding models |
The following model families are supported for the AWS Bedrock (EMBEDDING_PROVIDER=aws),
Model Family | Example Model ID | Request Format |
|---|---|---|
Amazon Titan |
|
|
Cohere |
|
|
Set the model using the EMBEDDING_MODEL_ID environment variable (default: cohere.embed-english-v3).
When using OpenAI (EMBEDDING_PROVIDER=openai), configure the model using EMBEDDING_MODEL_ID and provide your API key via OPENAI_API_KEY.
If the OpenAI Embeddings API is not hosted on the default OpenAI API address, OPENAI_BASE_URL should be set, for example if OpenAI API is hosted on Azure.
Pre-compute embeddings for batch indexing
Calculate similarity between two texts
Store embeddings in external systems
Debug and validate embedding quality
Returns detailed health status with individual dependency checks.
GET /components/__health
curl https://api.example.com/components/__health
{
"status": "OK",
"healthChecks": [
{
"status": "OK",
"severity": "HIGH",
"name": "OpenSearch Serverless Vector Store Healthcheck",
"type": "opensearch-serverless",
"impact": "Vector search and retrieval operations will fail if AOSS is unreachable.",
"message": "AOSS index 'my-index' reachable",
"description": "Validates AOSS connectivity and permissions via lightweight _search(size=0).",
"troubleshooting":
"https://api.example.com/__health/__trouble#opensearch-serverless-health"
}
]
}Status | Meaning |
|---|---|
OK | All checks passed |
WARNING | Non-critical issues detected |
ERROR | Critical dependency failed |
Severity | Impact |
|---|---|
HIGH | Service fails to function |
MEDIUM | Partial functionality |
LOW | Minor feature affected |
Good To Go - a quick binary health check for load balancers and probes.
GET /components/__gtg
Parameter | Type | Default | Details |
|---|---|---|---|
cache | boolean | true | true = fast cached check, false = fresh check |
# Fast check (for frequent polling) curl "https://api.example.com/components/__gtg"# Fresh check (for deployment verification) curl "https://api.example.com/components/__gtg?cache=false"
HTTP 200
{"gtg":"OK"}HTTP 503
{"gtg":"UNAVAILABLE"}Returns service metadata and version information.
GET /components/__about
curl https://api.example.com/components/__about
{
"description": "GraphRAG Components API, Embeddings and vector search endpoints.",
"version": "1.0.0-rc2",
"versionInfo": {
"jdkVersion": "21"
},
"additionalInfo": {
"Implementation-Title": "graphrag-components",
"Spring-Boot-Version": "3.5.6"
}
}Returns build manifest details.
GET /components/manifest-info
curl https://api.example.com/components/manifest-info