Create a rtemis.llm Agent

Usage

create_agent(
  llmconfig,
  system_prompt = SYSTEM_PROMPT_DEFAULT,
  use_memory = TRUE,
  tools = NULL,
  max_tool_rounds = 3L,
  output_schema = NULL,
  name = NULL,
  allow_custom_tools = FALSE,
  logfile = NULL,
  verbosity = 1L
)

Arguments

llmconfig: LLMConfig: The LLM configuration to use. Create using one of config_Ollama, config_OpenAI, or config_Anthropic.
system_prompt: Optional character: The system prompt to use.
use_memory: Logical: Whether to use conversation memory.
tools: Optional list of Tool objects: The tools available to the agent.
max_tool_rounds: Integer: Maximum number of tool call rounds per query.
output_schema: Optional Schema: The output schema to enforce on the agent's response created using schema and field.
name: Optional character: The name of the agent.
allow_custom_tools: Logical: If TRUE, allow the agent to carry tools whose function_name is not in the package allowlist. Such tools must be built via create_custom_tool and supply their own function body. The caller vouches for that code: built-in package guarantees (allowlist + hash verification) do not apply to it. Defaults to FALSE.
logfile: Optional character: Path to the agent's security/audit log. If NULL, the package default ("rtemis.llm_security_log.jsonl" relative to getwd()) is used. Can be overridden per call via the logfile argument to generate.
verbosity: Integer: Verbosity level.

Value

Agent object

Author

EDG

Examples

# Requires Ollama server running and gemma4:e4b model available
if (FALSE) { # \dontrun{
  agent <- create_agent(
    config_Ollama(
      model_name = "gemma4:e4b",
      temperature = 0.2
    ),
    system_prompt = "You are professor of Trance at the Institute of Advanced Beat Studies.",
    use_memory = TRUE
  )
} # }