## Summary
Closes#41466
- Adds Anthropic Claude API support to the `prompt.js` AI helper,
detecting `claude-*` model names and routing to the Anthropic Messages
API (`https://api.anthropic.com/v1/messages`) with proper authentication
headers
- Switches both LLM calls in the query generator
(`get-llm-generated-sql.js`) from OpenAI models
(`gpt-4o-mini-2024-07-18` for schema filtration, `o3-mini-2025-01-31`
for SQL generation) to `claude-sonnet-4-6-20260218`
- Adds `anthropicSecret` config placeholder in `custom.js` (set via
`sails_custom__anthropicSecret` env var in production)
- Updates the query generator UI to reference "Anthropic" instead of
"OpenAI"
### Changes
| File | What changed |
|------|-------------|
| `website/api/helpers/ai/prompt.js` | Added Anthropic API branch
alongside existing OpenAI logic; system prompts use Anthropic's
top-level `system` parameter |
| `website/api/controllers/query-generator/get-llm-generated-sql.js` |
Both model references changed to `claude-sonnet-4-6-20260218` |
| `website/config/custom.js` | Added `anthropicSecret` config
placeholder |
| `website/views/pages/admin/query-generator.ejs` | Updated copy from
"OpenAI" to "Anthropic" |
### Deployment notes
The `sails_custom__anthropicSecret` environment variable must be set
with an Anthropic API key before deploying this change.
---
Built for
[mikermcneil](https://fleetdm.slack.com/archives/D0AFASLRHNU/p1773278374183489?thread_ts=1773271495.702919&cid=D0AFASLRHNU)
by [Kilo for Slack](https://kilo.ai/features/slack-integration)
---------
Co-authored-by: kiloconnect[bot] <240665456+kiloconnect[bot]@users.noreply.github.com>
Co-authored-by: Eric <[email protected]>
Related to: https://github.com/fleetdm/confidential/issues/9884
Changes:
- Updated the prompt in the `get-llm-generated-sql` action to include a
note about using wildcard characters when generating queries that use
the LIKE operator.
- Improved error handling in the `get-llm-generated-sql` action
Changes:
- Updated the has-query-generator-access policy to allow the policy to
be bypassed when a `sails.config.custom.enablePublicQueryGenerator`
value is set
- Updated the query generator to match the latest wireframes
- Updated `get-llm-generated-sql` to pass a system prompt to the schema
filtration prompt.
Changes:
- Updated the prompt helper to accept a new input: `systemPrompt`, if
provided and a supported baseModel is selected, the system prompt will
be included in the Open AI API request along with the prompt.
- Updated the models used by the query generator to improve the speed of
query generation.
Closes: #25465
Changes:
- Updated the query generator page to use socket requests to call the
`get-llm-generated-sql` action to bypass Heroku's 30-second request
timeout.
Changes:
- Updated the User model to have a new attribute: `canUseQueryGenerator`
a boolean attribute that will grant users access to the query generator
page on the website.
- Added a new policy: `has-query-generate-access`
- Moved the query generator page from the admin section of the website.
Note: before this change can be merged, the website's database will need
to be migrated to add the new column to the User table
2025-01-14 18:09:17 -06:00
Renamed from website/api/controllers/admin/get-llm-generated-sql.js (Browse further)
