Skip to main content
This Helm repository contains queries to produce output that the LangSmith UI does not currently support directly (e.g. obtaining trace counts for multiple organizations in a single query). This command takes a postgres connection string that contains an embedded name and password (which can be passed in from a call to a secrets manager) and executes a query from an input file. In the example below, we are using the pg_get_trace_counts_daily.sql input file in the support_queries/postgres directory.

​
Prerequisites

Ensure you have the following tools/items ready.
  1. kubectl
  2. PostgreSQL client
  3. PostgreSQL database connection:
    • Host
    • Port
    • Username
      • If using the bundled version, this is postgres
    • Password
      • If using the bundled version, this is postgres
    • Database name
      • If using the bundled version, this is postgres
  4. Connectivity to the PostgreSQL database from the machine you will be running the migration script on.
    • If you are using the bundled version, you may need to port forward the postgresql service to your local machine.
    • Run kubectl port-forward svc/langsmith-postgres 5432:5432 to port forward the postgresql service to your local machine.
  5. The script to run a support query

​
Running the query script

Run the following command to run the desired query: 
sh run_support_query_pg.sh <postgres_url> --input path/to/query.sql
For example, if you are using the bundled version with port-forwarding, the command might look like: 
sh run_support_query_pg.sh "postgres://postgres:postgres@localhost:5432/postgres" --input support_queries/pg_get_trace_counts_daily.sql
which will output the count of daily traces by workspace ID and organization ID. To extract this to a file add the flag --output path/to/file.csv

​
Export usage data

All export methods produce the same data: LangSmith trace counts, LangSmith Deployments node usage, and Agent Builder run counts across all workspaces and organizations.
The UI and API exports require both of the following:
  • The organization:manage permission.
  • The caller’s email must be listed in USAGE_EXPORT_ADMIN_EMAILS, or ORG_ADMINS_INSTALLATION_USAGE_EXPORT_ENABLED must be set to true.
USAGE_EXPORT_ADMIN_EMAILS='["admin@example.com", "admin2@example.com"]'
ORG_ADMINS_INSTALLATION_USAGE_EXPORT_ENABLED=true
  1. Navigate to Settings > Usage and billing > Usage export.
  2. Click Export usage data.
  3. A ZIP file containing all usage data will download.

​
Export via API

If you prefer to export usage data programmatically, you can call the export API endpoint directly. 
curl -OJ \
  -H "X-API-Key: <your_api_key>" \
  https://<langsmith_url>/api/v1/orgs/current/usage/backfill-export

​
Export via SQL scripts

You can also run SQL scripts directly against your PostgreSQL database to export usage data. This requires database access credentials — no application-level permissions apply. To export trace usage (requires Helm chart version 0.11.4 or later): 
sh run_support_query_pg.sh <postgres_url> \
  --input support_queries/postgres/pg_usage_traces_full_export.sql \
  --output ls_export.csv
To export node usage (requires Helm chart version 0.11.4 or later): 
sh run_support_query_pg.sh <postgres_url> \
  --input support_queries/postgres/pg_usage_nodes_full_export.sql \
  --output lgp_export.csv
To export Agent Builder run counts (requires Helm chart version 0.13.25 or later): 
sh run_support_query_pg.sh <postgres_url> \
  --input support_queries/postgres/pg_usage_agent_builder_full_export.sql \
  --output ab_export.csv
Edit this page on GitHub or file an issue.
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.