Skip to contents

Create an ellmer chat pre-wired with EBI OLS4 MCP tools

Source: R/ols4_mcp.R
ols4_chat.Rd

Calls ols4_mcp_tools to retrieve the 12 OLS4 search tools, creates an ellmer chat object via llm_chat, and registers all tools on the chat so the LLM can invoke them during structured calls.

Usage

ols4_chat(
  provider = "anthropic",
  model = "claude-sonnet-4-5",
  temperature = 0,
  url = "https://www.ebi.ac.uk/ols4/api/mcp",
  ...
)

Arguments

provider

character(1) LLM provider; see llm_env_var. Defaults to "anthropic".

model

character(1) model identifier. Defaults to "claude-sonnet-4-5".

temperature

numeric(1) sampling temperature passed to the LLM. Defaults to 0 for deterministic, reproducible output.

url

character(1) OLS4 MCP endpoint URL passed to ols4_mcp_tools.

...

additional arguments passed to llm_chat.

Value

an ellmer Chat object with all OLS4 tools registered.

Note

Starting the mcp-remote bridge takes a few seconds. Reuse the returned object across multiple calls rather than creating a new one each time.

Examples

if (interactive()) {
    ch <- ols4_chat()
    map_concepts("chromatin accessibility and histone modification", chat = ch)
}