Anthropic Provider for StackQL Available
The anthropic provider for stackql is now available in the dev stackql provider registry. The anthropic provider for stackql includes services for interacting with Claude models via the Messages API. To get started download stackql, set the ANTHROPIC_API_KEY environment variable and use the dev registry as shown here:
export DEV_REG="{ \"url\": \"https://registry-dev.stackql.app/providers\" }"
./stackql --registry="${DEV_REG}" shell
Then pull the anthropic provider using:
REGISTRY PULL anthropic;
Now you can run some queries. Here's an example running inference against Claude using the Messages API - the request parameters are supplied in the WHERE clause and the response is projected using JSON_EXTRACT:
SELECT
id,
model,
stop_reason,
JSON_EXTRACT(content, '$[0].text') AS assistant_message,
JSON_EXTRACT(usage, '$.output_tokens') AS output_tokens
FROM anthropic.messages.messages
WHERE model = 'claude-sonnet-5'
AND max_tokens = 1024
AND messages = '[{"role": "user", "content": "What is StackQL?"}]';
|------------------------------|-----------------|-------------|-------------------------------------------------------------|---------------|
| id | model | stop_reason | assistant_message | output_tokens |
|------------------------------|-----------------|-------------|-------------------------------------------------------------|---------------|
| msg_01MLTLVY6XCTT2cNBeFeJzfj | claude-sonnet-5 | end_turn | StackQL is a SQL-based framework that lets you query and | 48 |
| | | | manage cloud and SaaS resources using familiar SQL syntax. | |
|------------------------------|-----------------|-------------|-------------------------------------------------------------|---------------|
You can also discover which models are available and their capabilities using the vw_model_capabilities view:
SELECT id, display_name, thinking, adaptive, xhigh, max_input_tokens, max_tokens
FROM anthropic.models.vw_model_capabilities;
|-------------------|-----------------|----------|----------|-------|------------------|------------|
| id | display_name | thinking | adaptive | xhigh | max_input_tokens | max_tokens |
|-------------------|-----------------|----------|----------|-------|------------------|------------|
| claude-sonnet-5 | Claude Sonnet 5 | 1 | 1 | 1 | 200000 | 64000 |
| claude-opus-4-8 | Claude Opus 4.8 | 1 | 1 | 1 | 200000 | 32000 |
| claude-haiku-4-5 | Claude Haiku 4.5| 1 | 0 | 0 | 200000 | 64000 |
|-------------------|-----------------|----------|----------|-------|------------------|------------|
Like other language models, Claude's responses are stochastic, so you'll get slightly different responses each time you query.
Let us know what you think! ⭐ us on GitHub.