View usage information

Use /stats for historical usage and /context for the current session's context window.

/stats

/stats

The stats panel shows information such as:

• Total sessions.
• Total tokens.
• Favorite model.
• Active days and streaks.
• Token usage by model.

Inside the panel:

• Tab switches Overview / Models.
• r cycles date ranges.
• Ctrl+S copies the stats view.

/context

/context

It shows the current context usage, including system prompt, tool schemas, MCP tools, custom agents, memory files, skills, message history, and free space.

Use it to understand why a session is getting large or what consumes the most tokens.

Why /context can be slow

/context does more than read a cached number. It approximates the real API request by:

1. Applying the same message transforms used before API calls, including compact boundaries and microcompact.
2. Building the effective system prompt.
3. Counting tokens for system prompt, tools, MCP tools, agents, memory, skills, and messages.
4. Prefering provider token-counting APIs, then auxiliary token counting or local estimation as fallback.
5. Producing a categorized breakdown.

This can involve multiple token counting requests and scans of tools, skills, agents, and memory.

Which command to use

• Historical usage: /stats.
• Current context size: /context.
• Recent API failure details: /doctor api.
• Current model route: /model show.