Anthropic LLM¶

Runs prompts against Anthropic Claude models via the Salt LLM service. It supports dynamic prompt composition, model auto-discovery with graceful fallback, temperature control, and max token limits. Designed for general text generation tasks like summarization, Q&A, drafting, and analysis.

Usage¶

Use this node when you need text responses from Anthropic models. Provide a system prompt to control tone and rules, a main prompt (which can include placeholders), and optionally connect up to four string inputs (e.g., retrieved knowledgebase content) to dynamically fill the prompt. Select a model from the dropdown; if the live model list isn't available, a curated fallback list is used. The node returns a single string with the model's response.

Inputs¶

Field	Required	Type	Description	Example
model	True	STRING (select)	Anthropic model to use. The list is fetched from the LLM service; if unavailable, a fallback list of Claude models is provided.	claude-3-5-sonnet-20241022
system_prompt	True	STRING	Instructions and rules for the assistant’s behavior, tone, and format.	You are a helpful assistant. Answer concisely and cite sources when possible.
prompt	True	DYNAMIC_STRING	User-facing request. Supports placeholders for connected inputs using {{input_1}}..{{input_4}}.	Summarize the following research notes in 5 bullet points: {{input_1}}
temperature	True	FLOAT	Controls creativity and variability. Lower = more deterministic; higher = more diverse.	0.5
max_tokens	True	INT	Maximum tokens to generate. Influences response length.	1024
input_1	False	STRING	Optional contextual input. Refer to it in the prompt as {{input_1}}.	Long-form notes or retrieved context text
input_2	False	STRING	Optional contextual input. Refer to it in the prompt as {{input_2}}.	Additional document text
input_3	False	STRING	Optional contextual input. Refer to it in the prompt as {{input_3}}.	Knowledgebase snippet
input_4	False	STRING	Optional contextual input. Refer to it in the prompt as {{input_4}}.	User profile or constraints

Outputs¶

Field	Type	Description	Example
Output	STRING	The generated text from the selected Anthropic model.	Here is a concise, 5-bullet summary of the provided notes...

Important Notes¶

Model selection: The node queries the live Anthropic model list; if unavailable, it falls back to known Claude models (e.g., claude-3-7-sonnet-20250219, claude-3-5-sonnet-20241022, claude-3-haiku-20240307).
Dynamic placeholders: Use {{input_1}} to {{input_4}} inside the prompt to inject optional inputs.
Response length: max_tokens bounds the completion length and may be trimmed by the service if too large.
Timeouts: The node uses a request timeout (default 90s). Long-running requests may need a higher timeout configured upstream.
Required configuration: The Salt LLM service must be configured with valid Anthropic API credentials for requests to succeed.

Troubleshooting¶

Model ID not found: Select a model from the dropdown. If you pasted a name, ensure it matches an available or fallback model.
Empty or very short output: Increase max_tokens or reduce temperature; verify your prompt and system prompt are not empty and that placeholders are populated.
Request timed out: Reduce prompt/context size, lower max_tokens, or adjust the service timeout configuration.
Inaccurate or off-topic responses: Refine the system_prompt with clearer rules and add more specific examples/context via input_1..input_4.
Service error from LLM: Verify Anthropic credentials and connectivity in the Salt LLM service; try another model from the list.