Skip to content

CostGuard and where your data goes

This is the lesson that answers the two questions almost everyone has on their first week with Clawless. The first is “what if I accidentally rack up a giant bill?” The answer is CostGuard: a software safety net you configure once with a monthly cap, a warn threshold, and a hard-stop behavior. Pay-as-you-go API calls count toward the cap; OAuth (Codex through your ChatGPT subscription) and local models do not. The dock row indicator goes yellow at the warn threshold, red at 100%. The second question is “who is actually seeing my conversations?” The answer is short. The data path is your computer, the AI provider you picked, your computer. There is no Clawless server holding your conversations. The one server we run is a license server, and it only sees license state. The lesson also covers what survives if Clawless ever shut down (your data; it is on your computer) and the four-party trust model worth applying to any tool, skill, or integration you install.

Prerequisites: lesson 3, API keys and the OAuth path, where you connected at least one AI provider and met Codex. You need Clawless installed with that provider connected; ideally you have run a few real conversations so the Usage dashboard has actual numbers to look at. The practice is a 20-minute setup (configure CostGuard, open the Usage dashboard, find the privacy policy).

  • Configure a CostGuard cap, warn threshold, and hard-stop behavior in Settings, Budget, and predict how the dock row indicator will signal you across the period
  • Distinguish what counts toward the cap (pay-as-you-go API keys) from what does not (OAuth via Codex, local models)
  • Trace the data path of a single message from your keyboard to the agent’s reply, and identify the one server in the path
  • Identify the three places your stuff lives (your computer, the AI provider, our license server) and what each one holds
  • Apply the four-party trust model to evaluate any new tool, skill, or integration before installing it
  • Read time: about 12 minutes
  • Practice time: about 20 minutes (set a monthly cap that feels comfortable, pick warn and hard-stop, save; open the Usage dashboard for the past week; find and skim the in-app privacy policy)
  • Difficulty: intro