Reproducible Language-Model Agents for Research

---

Documentation for package ‘LLMRagent’ version 0.8.0

• DESCRIPTION file
• User guides, package vignettes and other documentation
• Package NEWS
• Package README

Help Pages

add_edge	Add an edge to a workflow
add_node	Add a node to a workflow
Agent	The Agent class
agent	Create an agent
Agent-class	The Agent class
agent_as_tool	Expose an agent as a tool for other agents
agent_calibrate	Calibrate LLM/agent labels for valid downstream inference
agent_experiment	Run a factorial agent experiment
agent_manifest	Build the study manifest for a run
agent_pipeline	Run input through a chain of agents
agent_population	Build a population of agents
agent_robustness	Run a robustness battery
agent_run	Convert an LLMRagent result to a unified run object
agent_tool	Define a governed tool
agent_workflow	Build an agent workflow (a small, explicit graph)
approve_tool_call	Approve, reject, or edit a pending tool call
archive_agent_study	Seal an agent study to a directory
as.character.persona_frame	A persona as an auditable research object
as_agent_run	Convert an LLMRagent result to a unified run object
as_llmrcontent_validation	Build a validation frame for LLMRcontent
as_tibble.agent_run	Convert an LLMRagent result to a unified run object
attach_calibration	Attach a calibration to an agent run
budget	Spending and effort limits for an agent
check_state_leakage	Detect shared state across experiment cells
collect_measures	Collect measures over a society
contamination_report	Flag shared agent instances across a population
conversation	Run a multi-agent conversation
debate	Structured debate between two agents
deliberate	Group deliberation with a recorded vote
diagnostics.Agent	Machine-readable diagnostics for an agent run
diagnostics.agent_calibration	Machine-readable diagnostics for a calibration
diagnostics.agent_experiment	Machine-readable diagnostics for an agent experiment
diagnostics.agent_run	Machine-readable diagnostics for an agent run
diagnostics.persona_audit	Machine-readable diagnostics for a persona audit
exposure_matrix	Exposure matrix: who could see whom
focus_group	A moderated focus group
fork_workflow	Fork a workflow run at a checkpoint
guardrail	Define a guardrail
guardrails	Collect guardrails
hash_persona	Hash a persona
hash_tool_spec	Hash a tool's full specification
hash_workflow	Hash a run's control flow (the workflow)
human_gate	Mark a point or a tool as requiring human approval
interview	A semi-structured interview
llmragent-methods	LLMR-family methods for LLMRagent run objects
llm_claim_lint	Assert (or scope) prose against a run's claim type
load_agent	Load an agent from disk
mark_claim_type	Mark the kind of claim a run can support
mcp_tools	Expose MCP server tools to an agent, under governance
memory	Agent memory policies
memory_buffer	Agent memory policies
memory_recall	Agent memory policies
memory_summary	Agent memory policies
persona_audit	Audit persona briefs for essentializing language and caricature
persona_frame	A persona as an auditable research object
persona_variants	Vary a persona along named dimensions
print.agent_population	Build a population of agents
print.persona_audit	Audit persona briefs for essentializing language and caricature
print.persona_frame	A persona as an auditable research object
print.persona_set	Vary a persona along named dimensions
print.society	Assemble a society from a population, a network, and measures
replay_run	Replay a workflow run, verifying state hashes
report.Agent	Draft a methods-section report for an agent run
report.agent_experiment	Draft a short report for an agent experiment
report.agent_run	Draft a methods-section report for an agent run
report.society	Draft a disciplined report for a society
reset.Agent	Clear an agent's memory
resume_run	Resume a paused run after a tool-approval decision
resume_workflow	Resume a paused or failed workflow run
robustness-axes	Robustness perturbation axes
run_workflow	Run a workflow
sandbox_tool	Define a confined (sandboxed) tool
save_agent	Save an agent to disk
society	Assemble a society from a population, a network, and measures
step_interaction	Advance one interaction round
think_harder	Work a hard problem with one strong model and many cheap ones
vary_models	Robustness perturbation axes
vary_option_order	Robustness perturbation axes
vary_persona	Robustness perturbation axes
vary_prompt	Robustness perturbation axes
vary_temperature	Robustness perturbation axes
view_run	View a run as a self-contained HTML inspector
workflow_from_pipeline	Express an agent pipeline as a workflow