Skip to content
Free · no account · no key

You hit your Claude Code limit on Thursday. Again.

The limit tells you that you ran out. It never tells you what on. One free command reads the sessions already sitting on your disk and prices your week: where the tokens went, and which habit sent them there. It reads your transcripts, never your code.

Reads ~/.claude/projects. Prints to your terminal. Sends nothing without asking you first.

Sample output. The command prints what the transcripts still on disk cost you, what your standing config charges before you have typed anything, and how much of the bill went to context you carried rather than work you asked for.
What you get

Six answers you cannot currently get.

Each one is a habit with a number attached, read straight out of what your sessions recorded. There is no score and no grade.

  1. Where the money went

    The share of the bill spent re-sending context, which no invoice breaks out.

  2. Files you paid to read twice

    How often the agent re-opened a file it already had in context.

  3. Your own degradation point

    Where your re-reads and tool errors start climbing, fitted from your sessions.

  4. What your config costs per turn

    What your CLAUDE.md, skills and MCP servers charge before you state the task.

  5. Tasks that went to too big a model

    Sessions routed above what the work needed, priced against the right tier.

  6. What each subagent spawn carries

    The prefix tokens every spawn drags along, and the monthly total.

And then what

Three plans, written by your own agent.

It hands those numbers to the agent already on your machine and asks for three changes, ranked by what they save. Nothing is uploaded and no model of ours is involved. The run spends your own limit, and it tells you the cost before it does.

Sample output. Each plan quotes the field and figure it came from, says what the change is worth, and names the number that should move if it worked.
  • Your agent, not ours. It shells out to claude -p or codex exec, whichever you already have.
  • Findings only. The agent is handed the measured numbers. Your raw transcripts never reach it.
  • No tools, no repo access. It gets text and returns text, so read-only is structural rather than promised.
  • It remembers. Plans are kept locally, so a later run says whether the thing you changed moved.

Want to see exactly what your agent would be asked before it is asked? npx @promptster/cc-audit --print-prompt prints the prompt and invokes nothing.

What it does and does not touch

It never reads your source code.

A tool that reads your dev sessions has to be boring about privacy, so here is exactly what it looks at and what it leaves alone.

  • It never reads your source code. The analysis runs on token accounting, tool calls, and file paths. Not the contents of the files you edited. If you are working on an employer's codebase, nothing about that codebase is examined.
  • It runs on your machine. It reads the session transcripts Claude Code already writes to ~/.claude/projects. There is no account, no key, no sign-up, and no daemon left running.
  • Nothing happens without a yes. The measurement is local and prints here. In a terminal the run then asks before anything else: once to write the plans on your own agent, once to publish a shareable link. That second yes also switches on data sharing, which only cc-audit capture --off switches back. No is a complete answer to both. The --judge and --open flags and fix each send something too, and each is a flag you type. Your prompt gists stay on your machine.
  • Nothing is left behind. npx it once and it is gone. fix writes proposed patches under ./.cc-audit/ and never applies them itself.

The exact contents of each of those three flags, how long anything they send is kept, and how to have it deleted are written out on the CLI data policy page.

How it works

Three commands, and you only need the first.

Steps two and three are optional and cost you nothing to skip. The first one measures locally, then asks before it writes the plans or publishes anything.

  1. Run it

    No install, no account, no key. It reads the session transcripts Claude Code already writes to ~/.claude/projects, prints the report, then asks two things: whether to write the plans on your own agent, and whether to publish a shareable link. No to both leaves everything on your machine.

  2. Right-size your modelsoptional

    Optional, and the flag is the consent. Sends a task gist plus metadata — never your code — to the hosted judge, then prints which sessions were over-modelled and what the cheaper tier would have cost.

  3. Turn it into patchesoptional

    Optional. Writes reviewable patches under ./.cc-audit/ — local model-pin edits plus a hosted CLAUDE.md trim. It never applies them itself.

Running this across a whole engineering org asks a different question, and one laptop's logs cannot answer it. That one is Promptster for Teams, which is a paid product.

Want this without running it yourself?

The command is the product and it stays free. A hosted version that keeps the same numbers up to date between runs, instead of you remembering to re-run it, is the thing being built next.

One email when it ships. No sequence, no newsletter. If it hasn't shipped within a year, your address is deleted — see the CLI data policy.

Frequently asked questions

  • Is it actually free?
    Yes. It is a free command with no account, no key, no card, and no seat count. There is nothing to sign up for and no trial to expire.
  • Does it work if I use Codex or Cursor instead of Claude Code?
    Not yet. It reads the session transcripts Claude Code writes to ~/.claude/projects, so today it is Claude Code only.
  • Will this tell me how to stop hitting my weekly limit?
    It tells you what spent it, which is the part you cannot currently see. Most of what it finds is habit rather than volume: context you carried and never used, files re-read after they were already in the window, and tasks handed to a model several tiers larger than the work needed.
  • Does it send my code anywhere?
    No. It never reads the contents of your files: the analysis runs on token accounting, tool calls, and file paths. The measurement itself uploads nothing. Anything that does leave is either a flag you typed or a prompt you answered yes to, and the run tells you what it covers before you answer.
  • What is the difference between this and Promptster for Teams?
    This shows one developer their own habits on their own machine. Promptster for Teams answers a different question for an engineering organization, across many engineers, and is a paid product. The free command does not require or lead to it.
  • Do my sessions get used to build benchmarks for other companies?
    No. What can be shared is a de-identified summary, and it is shared only after you say yes to the prompt that describes it or type one of the flags. Even then, your individual sessions are not turned into org-level comparisons for anyone else.