DeepSeek
CertifiedTasks that call DeepSeek chat models for conversational or schema-guided JSON responses.
Use DeepSeek ChatCompletion to orchestrate role-based conversations or JSON-only replies within Kestra Flows. Requires apiKey, modelName, and messages; defaults baseUrl to https://api.deepseek.com/v1 for OpenAI-compatible routing. Add jsonResponseSchema to enable JSON Mode with a prepended schema reminder; DeepSeek treats the schema as guidance rather than server-side validation.
DeepSeek
Tasks that call DeepSeek chat models for conversational or schema-guided JSON responses.
Use DeepSeek ChatCompletion to orchestrate role-based conversations or JSON-only replies within Kestra Flows. Requires apiKey, modelName, and messages; defaults baseUrl to https://api.deepseek.com/v1 for OpenAI-compatible routing. Add jsonResponseSchema to enable JSON Mode with a prepended schema reminder; DeepSeek treats the schema as guidance rather than server-side validation.
How to use the DeepSeek plugin
Call DeepSeek chat models for conversational completions and structured JSON responses.
Authentication
Set apiKey to your DeepSeek API key. Store it in a secret. The baseUrl defaults to https://api.deepseek.com/v1 — override it if you are using a self-hosted or OpenAI-compatible proxy.
Tasks
ChatCompletion sends a list of messages to a DeepSeek model and returns the response. Set jsonResponseSchema to enable JSON mode — the model will return a structured JSON response guided by the schema you provide. Note that DeepSeek treats the schema as a prompt-level hint rather than performing server-side validation, so downstream parsing should still handle unexpected output shapes.
