Connectors & Bindings

Anything Graph connects your agents to live data without copying it into a new warehouse. Register a source in your profile, introspect its schema through MCP, then author bindings that map tables, collections, or API paths to entities in an ontology playbook — with credentials kept in .env, never in playbooks.

Pick a connector below for step-by-step setup, or start with Connect your data to add a source with the CLI.

Currently supported adapters

The adapters shipped in the repo today are the adapter-* crates in the ag-cli/crates directory:

Crate	Profile adapter	Typical sources
adapter-sql	sql	PostgreSQL (via sqlx)
adapter-csv	csv	Local CSV and flat files
adapter-soql	soql	Salesforce REST API
adapter-mongodb	mongodb	MongoDB collections
adapter-rest	rest	HTTP JSON APIs with OpenAPI specs

Each adapter is a Rust crate implementing the shared adapter-core DataAdapter trait. More connectors are on the roadmap — same playbook model, different backends.

All adapters (including planned)

Adapter	Profile adapter	Typical sources	Introspect (MCP)	Status
SQL	sql	PostgreSQL (via sqlx)	Tables, columns, foreign keys	Available
CSV	csv	Local CSV and flat files	Column headers from file	Available
SOQL	soql	Salesforce REST API	Describe objects & fields	Available
MySQL	mysql	MySQL, MariaDB	Tables, columns, keys	Available
SQL Server	mssql	Microsoft SQL Server, Azure SQL	Tables, columns, keys	Available
MongoDB	mongodb	MongoDB collections	Collections & field samples	Available
REST / OpenAPI	rest	HTTP JSON APIs with OpenAPI specs	Paths, parameters, response shapes	Available
BigQuery	bigquery	Google BigQuery datasets	Datasets, tables, columns	Planned
Snowflake	snowflake	Snowflake warehouses	Schemas, tables, columns	Planned
Databricks	databricks	Databricks SQL / Unity Catalog	Catalogs, schemas, tables	Planned
Elasticsearch	elasticsearch	Elasticsearch, OpenSearch indices	Index mappings	Planned
S3 / Parquet	s3	Amazon S3, object storage, Parquet files	Prefixes, columns from Parquet	Planned
GraphQL	graphql	GraphQL endpoints	Schema introspection	Planned
Google Sheets	google_sheets	Google Sheets spreadsheets	Sheet tabs & column headers	Planned
HubSpot	hubspot	HubSpot CRM objects	Object properties	Planned

Binding patterns

• mysql / mssql — same YAML shape as Postgres (from, fields, subject_link_column); SQL dialect is compiled automatically
• mongodb — set from to a collection name; runtime compiles find: / count: operations
• rest — set from to an API path (e.g. /users); runtime compiles GET request templates; profile needs base_url

Binding file suffix often matches the source key (.postgres, .mysql, .mongodb, .rest, …). The adapter type comes from the profile entry referenced by source_id.

Configure a connector

Register sources in profiles/local.yaml and set credentials in .env. See Connect your data for profile examples and environment variables.