I kept approving Claude Code commands I didn't understand, so I built clear-caption
Description and use
The clear-caption plugin adds plain‑English captions to Claude Code commands, enabling quick understanding and highlighting risky actions.
Problem
Approving complex Claude Code commands without understanding them leads to accidental destructive actions.
Outcome
Clear captions reduce accidental destructive actions by providing concise, readable summaries of each command, with explicit warnings for high‑risk operations.
Workflow
- 1. Install clear-caption via /plugin install clear-caption
- 2. Enable captions with /caption all (default) or /caption risky
- 3. For each Claude Code call, read the caption above the command
- 4. Approve or reject based on caption and warning
- 5. Execute command
Evidence
I approve dozens of commands from Claude Code every day.
Now, before anything runs, Claude tells me in plain English what it is about to do.
Some commands can wipe out your work in one keystroke. Those get a ⚠ in the caption, even when the description sounds harmless.
On testing: edit captions landed on 57 of 57 edits across Opus, Sonnet, and Haiku. The average cost on the Opus test tasks was the same with captions on or off, for about 3% more output tokens.
- Record date: 2026-10-20T22:41:04.490Z