Module agent 

Agent module for interactive AI-powered CLI assistance

This module provides an agent layer using the Rig library that allows users to interact with the CLI through natural language conversations.

Features

• Conversation History: Maintains context across multiple turns
• Automatic Compaction: Compresses old history when token count exceeds threshold
• Tool Tracking: Records tool calls for better context preservation

Usage

# Interactive mode
sync-ctl chat

# With specific provider
sync-ctl chat --provider openai --model gpt-5.2

# Single query
sync-ctl chat --query "What security issues does this project have?"

Interactive Commands

• /model - Switch to a different AI model
• /provider - Switch provider (prompts for API key if needed)
• /help - Show available commands
• /clear - Clear conversation history
• /exit - Exit the chat

Modules

commands

Slash command definitions and interactive command picker

compact

Context compaction module (forge-inspired)

history

Conversation history management with forge-style compaction support

ide

IDE Integration Module

persistence

Session persistence for conversation history

prompts

Embedded prompts for the Syncable agent

session

Interactive chat session with /model and /provider commands

tools

Agent tools using Rig’s Tool trait

ui

Terminal UI module for agent interactions

Structs

AgentState

Agent state for AG-UI state synchronization

ConversationState

Conversation state for AG-UI

TokenUsageState

Token usage state for AG-UI

Enums

AgentError

Error types for the agent

ProviderType

Provider type for the agent

Functions

run_agent_server

Run the agent as a dedicated AG-UI server (headless mode for containers/deployments).

run_interactive

Run the agent in interactive mode with custom REPL supporting /model and /provider commands

run_query

Run a single query and return the response Note: event_bridge is accepted for API consistency but not used in single-query mode

Type Aliases

AgentResult