Claude Prompt Cache Control — Development skill for Claude Code

prompt-cache-control

A prompt-cache meter above the Claude Code prompt. Every request Claude makes reports how much of its prompt the cache served, how much it wrote and how much went uncached; this mod keeps those numbers per request and per turn, counts down to the moment the cache lapses and tells you what to do about it: keep going, `/compact` or `/clear`.

cache ██████████ 98% read 80k · wrote 1k · new 300 ⏱ 3:41 5m · warm: keep going
cache ░░░░░░░░░░  0% read 0 · wrote 52k · new 300 ⏱ 4:58 5m · cache missed: model changed (…)
cache ██████████ 98% read 150k · wrote 1k · new 300 ⏱ 0:00 5m · expired: the next message rewrites 151k tokens. /compact first, or /clear if the task is done

`/cache` opens a pane: the time left with a solid bar that shrinks as the cache runs out (green, yellow below 40% of the lifetime, red from the warning threshold), a stacked read / wrote / new bar for the last request, and a colour-coded table with one row per turn. Bars are filled cells and columns have fixed widths with no-break spaces, so the terminal and the Desktop (HTML) pane render the same.

Clear button

The band and the `/cache` pane carry a **Clear** button (hotkey `c` once the band or pane has focus: ctrl+x tab for the band). Pressing it:

1. reads this session's transcript (~/.claude/projects//.jsonl), keeps the user and assistant text and one line per tool call, and asks haiku for a brief with fixed headings: Goal, Current state, Decisions made, Files and commands touched, Open items, Next step;
2. runs /clear, so the next request starts from an empty, cheap prefix;
3. submits the brief to the fresh conversation as your own first prompt, framed to ask for a one-line acknowledgement and then wait.

If the model gives no brief, a fallback with your last prompts and a `claude --resume ` line is sent instead. If `/clear` or the submit is refused, the brief is put in the prompt box for you to send. The button is hidden while