Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

47 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

/cache-stats

/cache-stats is a lightweight TUI plugin for opencode that displays the active session's cache statistics in a popup dialog. It provides the overall cache hit rate, a breakdown by model and subagent, and the associated costs, making it easy to understand cache performance at a glance.

Run the /cache-stats command (or the cache-stats.show command) inside a session to open the dialog.

Summary and models Totals breakdown Subagents
Session summary and models Grouped totals Subagent list

What it shows

The dialog shows:

  • The overall cache hit percentage of the active session. A hit bar and a verdict word (Excellent, Good, Fair, or Poor) show the level.
  • A session summary: total calls, model count, subagent count, total tokens, and total cost.
  • A per-model breakdown with hit bars, hit percentages, and token totals.
  • Grouped totals for tokens, calls, and cost. The totals include percentages and averages per call.
  • A subagent list with the hit rate of each subagent and the model that each subagent used.

The plugin computes the cache hit rate with this formula:

hit = cache read / (input + cache read)

A cache write is a one-time priming cost. It is excluded from the hit ratio. The plugin also reports the number of API calls that read from the cache.

Install

  1. Copy the plugin into ~/.config/opencode/tui-plugins/.
  2. Register the plugin in ~/.config/opencode/tui.json:
{
  "$schema": "https://opencode.ai/tui.json",
  "plugin": ["./tui-plugins/cache-stats.tsx"]
}

The repo keeps a copy of tui.json as tui.json.example.

After you change tui.json, restart opencode. Then run the /cache-stats command inside a session.

Documentation

About

/cache-stats is a lightweight TUI plugin for opencode that displays the active session's cache statistics in a popup dialog. It provides the overall cache hit rate, a breakdown by model and subagent, and the associated costs, making it easy to understand cache performance at a glance.

Topics

Resources

Stars

14 stars

Watchers

0 watching

Forks

Contributors

Languages