Chat Export

Save your conversations with AI agents for future reference.

Manual Export

Click the export button in the chat header to export the current conversation.

Export Settings

Configure export options in Settings → Agent Client → Export:

Setting	Description
Export folder	Folder where chat exports will be saved (default: Agent Client)
Filename	Template for filenames. Use {date} and {time} as placeholders
Frontmatter tag	Tag added to exported notes (default: agent-client). Supports nested tags like projects/agent-client
Auto-export on new chat	Automatically export when starting a new chat
Auto-export on close chat	Automatically export when closing the chat view
Open note after export	Automatically open the exported note
Include images	Save images attached in messages and returned by tools (default: enabled)
Image location	Where to save images: Obsidian's attachment folder, custom folder, or embed as Base64
Custom image folder	Folder path for images when using custom location
Include audio	Save audio returned by tools (default: enabled)
Audio location	Where to save audio: Obsidian's attachment folder or a custom folder
Custom audio folder	Folder path for audio when using custom location

Export Format

Chats are exported as Markdown files with YAML frontmatter:

---
created: 2025-12-13T00:31:12
agentDisplayName: Claude Code
agentId: claude-agent-acp
session_id: f95b4847-cb9c-441a-9f0b-08eb243ff5dd
tags: [agent-client]  # Customizable in settings
---

# Claude Code

## 0:31:12 - User

@[[Agent Client Plugin]]
Summarize this, please.


---

## 0:31:16 - Assistant

### 🔧 Read File

**Locations**: `/Users/rait09/Documents/dev_vault/Agent Client Plugin.md:0`

**Status**: completed

## Summary: Agent Client Plugin for Obsidian

This is an Obsidian plugin that integrates AI coding agents (Claude Code, Codex, Gemini CLI) directly into your vault.
...

What Gets Exported

• Messages: Full conversation history with timestamps
• Images: Attached images (saved as files or embedded, based on settings)
• Tool calls: Tool name, locations, status, and diffs
• Tool results: Text output (fenced), images and audio (saved as attachments), resource links, and embedded resources; raw output when a tool returned no content
• Thinking: Agent's reasoning (as collapsible callouts)
• Plans: Task plans with status indicators
• Note mentions: Auto-mention and manual mentions in @[[note]] format

Use Cases

• Documentation: Keep records of conversations and decisions
• Learning: Review agent explanations later
• Sharing: Share solutions with others
• Debugging: Reference what the agent did for troubleshooting