Snowflake integration for AI agents
Query semantic views, ask Cortex Analyst, and run SQL
Connect Snowflake to Langdock and use it in chat, in agents and in workflows.
9 actions
Snowflake actions
Each action is one step an agent or workflow can run in Snowflake.
- Ask Cortex Analyst
Asks Snowflake Cortex Analyst a natural-language question against a semantic view, then runs the generated SQL.
- Describe semantic view
Describes the metrics, dimensions, facts, and relationships on a Snowflake semantic view.
- Execute SQL
Execute SQL query and return resulting data.
- List Cortex Search services
Lists Cortex Search services the current role can see. Use the qualifiedName with Query Cortex Search service (REST).
- List semantic views
Lists Snowflake semantic views the current role can see. Use before describing or querying a semantic view.
- Preview Cortex Search (SQL)
Runs Snowflake's SEARCH_PREVIEW SQL debug function against a Cortex Search service. Prefer Query Cortex Search service (REST) for the production REST API.
- Query Cortex Search service (REST)
Searches a Cortex Search service through the production REST API. Use this instead of Preview Cortex Search (SQL), which only runs Snowflake's SEARCH_PREVIEW debug function. Prefer a fully qualified name from List Cortex Search services.
- Query semantic view
Queries a Snowflake semantic view by metrics and dimensions. Do not mix facts and metrics in one query.
- Search schema
Searches for schemas or tables containing specific keywords.
Setup and permissions
An action can ask for your confirmation before it runs. In an agent, each action is set to “Always allow” or “Needs approval”.