aws-bedrock Embedding
aws-bedrock Embedding
Defines an embedding provider backed by the official AWS Bedrock InvokeModel API.
The aws-bedrock embedding resolves text to a vector through a configured, managed Bedrock embedding model, so a consumer such as binding-mcp-ext's semantic tool-search index can rank by cosine similarity against a hosted model instead of a locally-run one. Both Titan Text Embeddings and Cohere Embed request/response shapes are supported.
embeddings:
my_aws_bedrock_embedding:
type: aws-bedrock
options:
region: us-east-1
model: amazon.titan-embed-text-v2:0Configuration (* required)
options*
object
The aws-bedrock specific options.
embeddings:
my_aws_bedrock_embedding:
type: aws-bedrock
options:
region: us-east-1
model: amazon.titan-embed-text-v2:0options.region
string
AWS region where the Bedrock model is deployed. When omitted, resolved from the standard AWS region provider chain (environment variable, shared config/credentials file, or container/instance role).
options.model*
string
Bedrock model id to invoke, e.g. amazon.titan-embed-text-v2:0, cohere.embed-english-v3, or cohere.embed-multilingual-v3.

