> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Agent • AI Agent SDK
> Agent: Class reference for Agent
# Agent
> Defined in the [**agent**](../modules/agent) module.
AI Agent
```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#8B0000', 'primaryTextColor': '#fff', 'primaryBorderColor': '#710101', 'lineColor': '#189AB4', 'secondaryColor': '#189AB4', 'tertiaryColor': '#fff' }}}%%
graph LR
input["Input Data"] --> agent["Agent: Agent"]
agent --> output["Output Result"]
style agent fill:#8B0000,color:#fff
style input fill:#8B0000,color:#fff
style output fill:#8B0000,color:#fff
```
## Constructor
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
No description available.
## Methods
Create an Agent from a template.
Lazy-loaded StreamEventEmitter for real-time events (zero overhead when not used).
Allow setting stream\_emitter directly.
Instance method.
Instance method.
Return a fully independent copy of this agent for a gateway channel.
Get chat history (read-only access, use context managers for modifications).
Set chat history (updates the underlying async-safe state with lock).
AutoMemory instance for automatic memory extraction.
Instance method.
PolicyEngine instance for execution control.
Instance method.
OutputStyle instance for response formatting.
Instance method.
ThinkingBudget instance for extended thinking control.
Instance method.
Unified, provider-portable reasoning-effort level (Issue #4452).
Instance method.
Cumulative USD cost of all LLM calls in this agent run.
Summary of cost and token usage.
ContextManager instance for unified context management.
Set context manager directly.
Lazily initialize Rich Console only when needed AND verbose is True.
Lazily initialize SkillManager only when skills are accessed.
Get the XML prompt for available skills.
Structured reason the last tool-execution loop stopped.
Lazily generate agent ID when first accessed.
Safe display name that never returns None.
Root filesystem change-tracking at `project_path`.
Undo the last set of file changes.
Redo a previously undone set of file changes.
Get file diffs from autonomous execution.
Analyze prompt for autonomy signals.
Get recommended execution stage for prompt.
Run an autonomous task execution loop.
Async variant of run\_autonomous() for concurrent agent execution.
Programmatically hand off a task to another agent.
Run agent iteratively until output meets quality criteria.
Async version of run\_until().
Asynchronously hand off a task to another agent.
Register a verification hook for the current run.
Remove a verification hook by name. Returns True if one was removed.
Get tools available to this agent, filtered by plan\_mode if enabled.
Attach an MCP server to a live agent at runtime.
Disconnect and deregister a previously attached MCP server.
Re-derive the live toolset from `self.tools`.
Return the names of MCP servers currently attached at runtime.
Lazy-initialized RulesManager for persistent rules/instructions.
Get rules context for the current conversation.
Get memory context for the current conversation.
Get learning context for injection into system prompt.
Store content in memory.
Unified property to get the LLM model regardless of configuration type.
Get the unified retrieval configuration.
Lazy-loaded RAG instance for advanced retrieval with citations.
Retrieve context from knowledge without LLM generation.
Query knowledge and get a structured answer with citations.
Query knowledge using RAG pipeline with citations.
Chat with pre-retrieved context.
Synchronously close the agent and clean up resources.
Async version of close() for async context managers.
Returns True if the agent has been closed.
Plain-English answer to 'where does my code actually run?'.
## Source
`praisonaiagents/agent/agent.py` at line 247
***
## Related Documentation