4. OpenClaw TUI Interaction

TUI (Text User Interface) is OpenClaw's "command-line chat window". When it's inconvenient to open the web console on the RDK, or when you're connected via SSH remotely, the TUI allows you to complete most daily interactions directly in the terminal: conversation, model switching, status viewing, session management, skill invocation, etc.

1. How to Enter TUI

Execute in the terminal:

openclaw tui

If you have already started the OpenClaw gateway in another terminal (e.g., as a gateway/server), you can directly execute openclaw tui to enter the interaction interface.

For non-Docker versions, openclaw gateway does not need to be started manually and will start automatically on boot.

For Docker versions, you need to open a terminal first and manually run openclaw gateway to start the gateway.

2. What is the TUI Interface

You can think of TUI as a "chat box with commands":

• Top: Conversation history (your messages, model replies, tool/skill outputs)
• Bottom: Input box (enter questions or send slash commands)
• Send method: Press Enter after input to send

It is recommended to do two things when entering for the first time:

xxxxxxxxxx
/status

3. Slash Commands

Slash commands generally need to be sent as independent messages starting with /. They are used to control sessions and perform operations.

Additional "daily feel" suggestions:

• When sending a message with a slash command, it's safest to write only one slash command per message
• To quickly browse all commands, use /help or /commands

4. Session and Memory

In OpenClaw, these two concepts have clear divisions:

• Session: Manages "where this conversation has gotten to". When you use /new, /reset in TUI, you are essentially switching/reopening sessions.
• Memory: Manages "content that should be retained long-term across sessions". It is written to disk (Markdown) and retrieved back to the current session when needed.

4.1 Where are Session Files

Sessions are saved to disk in JSONL format. Common locations (based on actual installation path):

• Index: ~/.openclaw/agents/<agentId>/sessions/sessions.json
• Transcription: ~/.openclaw/agents/<agentId>/sessions/<SessionId>.jsonl

A more convenient way for actual troubleshooting is to execute first:

xxxxxxxxxx
openclaw status

It usually gives the store path and recent session information.

4.2 How to "Use" Memory

You can think of Memory as "having the robot take notes":

• memory/YYYY-MM-DD.md: Daily process records, suitable for recording progress, temporary decisions
• MEMORY.md: Long-term stable information (preferences, fixed rules, project constraints, etc.)

Practical suggestions:

• Want a conclusion to be usable across sessions: Just tell the robot "write this to MEMORY.md"
• Preparing to use /new to switch topics: First use /compact or write key decisions to Memory, then switch sessions

5. Quick Start Flow

1. Enter TUI: openclaw tui
2. Confirm status: /status
3. Select model: /models list → /model <number or name>
4. Daily conversation: Just enter questions and press Enter
5. Topic switch: /new
6. Context too long: /compact
7. Key conclusions to retain long-term: Have the robot write to Memory

6. Common Troubleshooting

6.1 Entered a command but no response

• Confirm the command is sent on its own line (don't mix with plain text)
• Use /help first to see which commands are supported in the current version

6.2 Feel "it's started talking nonsense, can't understand people" or being disturbed by history

• Start a new session: /new or /reset
• Check what context has been injected: /context detail

6.3 Responses getting slower

• Lower thinking depth: /think low or /think minimal
• Compact history: /compact