Module agent_traits
Source - CommandOutput
- Output from a
Command - ScopedExecutor
- Lightweight executor wrapper that applies a default working directory to forwarded commands.
- Command
- Commands that can be executed by the executor
Conceptually,
Shell allows any kind of input, and other commands enable more optimized
implementations. - CommandError
- ToolFeedback
- Feedback that can be given on a tool, i.e. with a human in the loop
- AgentContext
- Acts as the interface to the external world and manages messages for completion
- ExecutorExt
- Convenience methods for scoping executors without cloning them.
- MessageHistory
- A backend for the agent context. A default is provided for Arc<Mutex<Vec>>
- ToolExecutor
- A
ToolExecutor provides an interface for agents to interact with a system
in an isolated context.