Keep your cache warm.
Know the cost when it isn’t.

A mod for Claude Code’s prompt cache. Keep it warm while you take a break, or see the estimated rewrite cost before you send a cold prompt.

Free and open source. Claude Code 2.1.287+. Warming needs a one-hour cache and Claude Code left running. Pings cost tokens.

Before your break

Run /keepwarm 90m for a 90-minute window. After 50 idle minutes, Cache Tax sends a model request over the session to refresh its cache.

Each ping checks its usage. No cache reads, or writes reaching 10% of reads, stop the loop. The idle spending limit can skip further pings before the window ends, leaving the cache to go cold. /keepwarm off stops it too. An already-cold session waits for your next turn before pinging.

When you return cold

For a context of at least 50,000 tokens, the guard asks before your ordinary message goes out when its one-hour clock says cold. You see the estimated rewrite cost, then choose Send anyway or Don't send. Where nobody can answer, it stops the message once and the resend goes through.

By default, a detected cold write automatically arms a keepwarm window of at least three hours; pings may stop sooner at the idle spending limit. The Auto keepwarm option in /config turns automatic warming off. Run /cache-tax to see the window and this session’s cold-write tally.

Recorded in real sessions

One verified run on Fable 5.1: a ping at minute 50 read the cache, and the message sent at minute 62 read 156,886 tokens from cache. One run, one configuration.

Recorded warming receipt: last ping read 75k tokens, costing $0.02. Open full-size recording.
Sonnet 5, 16 September 2026: a 75k-token cache read at $0.02. This recording used the one-minute testing interval on 2.0.0. Its $0.45 rewrite estimate used an older price row; the current build prints $0.30 for the same context. Current versions show the receipt in the row above the prompt in terminal and Desktop sessions. A receipt for this session, not a universal ping price.
See the cold-send refusal
Recorded cold-send guard showing a $6.61 rewrite estimate. Open full-size recording.
Recorded on 2.1.1, before the 2.1.2 wording change. The estimate was $6.61; the subsequent reported cache write was $6.28. Current wording says “up to” the estimated token count. These are API-equivalent costs, not extra subscription charges.

Two commands to install.

Use Claude Code 2.1.287 or later. Mods are on by default; no early-access flag is needed.

claude plugin marketplace add karanb192/claude-code-mods
claude plugin install cache-tax@claude-code-mods

Restart Claude Code or run /reload-plugins in an open session. Then run /keepwarm 90m before your break. Plain /keepwarm arms six hours.

Check your cache lifetime. Warming uses a 50-minute timer. If your main cache defaults to five minutes, set promptCacheTtl to "1h" first.

Prefer a settings hook? The hook version warns by default, can refuse a cold send once, and provides a standalone status-line countdown. Warming is part of this mod.

Know what you are paying for.

  • A ping bills reads, writes, uncached input and output. Fork output is uncapped. The displayed dollars are API-equivalent estimates, not additional subscription charges. Quota savings have not been measured.
  • Prefix changes can break the cache before its timer expires. A warm read proves that ping hit the cache; it does not guarantee your next message will.
  • The mod makes model requests over your session and stores its settings. The README lists its hooks, engine calls and threat model, and the Claude Code mods catalogue shows the same scan.