References: CostGuard and where your data goes

Source material

The lesson is anchored on two chapters from the Clawless knowledge base, the single source of truth for Clawless behavior. The KB is reachable inside the app: click the Help button above the message box and ask the knowledge base by name; answers cite the docs chapter they come from. The product itself lives at clawless.ai.

Clawless KB chapters cited:

PRIMARY (the CostGuard half: settings, what counts, period rules,
visual cues, banner copy)
• Clawless Knowledge Base, "CostGuard" chapter
  Covers the three settings (monthly cap, warn threshold, hard-stop
  behavior), what counts toward the cap (pay-as-you-go vs OAuth vs
  local), period modes (rolling 30-day vs calendar-month), and the
  pre-release-vs-launch hedge on the enforcement layer.

PRIMARY (the privacy half: the data path, three places your stuff
lives, four-party trust model, what survives if Clawless shuts down)
• Clawless Knowledge Base, "What's Under the Hood" chapter
  Covers the desktop-app-plus-OpenClaw-engine architecture, the
  no-Clawless-server-in-the-data-path posture, the role of the license
  server (license state only), and the local-models option as the
  no-cloud guarantee.

Clawless documentation is open under Creative Commons Attribution-ShareAlike
(CC BY-SA 4.0). This lesson follows the KB's framing on cost limiting
and privacy posture, expressed in original prose for a Clawdemy reader
on their first week.

Going deeper inside Clawless

Two follow-on KB chapters that touch what CostGuard and the privacy posture do not cover directly.

• Clawless Knowledge Base, “Usage and Cost” chapter. The Usage dashboard in detail, including the per-provider, per-model, per-conversation breakdown that the lesson asks you to open. Useful when you want to understand where a specific dollar amount on your provider invoice came from.

• Clawless Knowledge Base, “Chat and Models” chapter. The model picker and the cost-per-model framing. Picking smaller models for routine work is the largest single lever on your BYOK spend, and that chapter is where to look when you want to make that decision deliberately.

Going deeper outside Clawless

Three outside sources for readers who want context on local-first design, AI provider data policies, and personal AI spending discipline.

• “Local-first software” (Ink and Switch, 2019). The original essay defining the local-first design principles that the Clawless privacy posture follows. Reads as a manifesto, but the seven principles in section 3 are a useful checklist for evaluating any tool that claims to be local-first. The “your data stays on your computer” framing in this lesson sits inside the broader local-first lineage this essay defines.

• Anthropic privacy policy and OpenAI privacy policy. Two of the major providers Clawless can route through. Read both at least once if any of your messages contain anything you would not want a model trained on. Both default to not training on API customer data, but the wording matters and the policies evolve, so the read-once-then-revisit cadence is the right one.

• The Verge: “AI assistants are getting expensive” (March 2024, illustrative). A general-press piece on what personal AI usage actually costs. Not Clawless-specific. Useful for calibration if you are wondering whether the dollar amounts you see in your Usage dashboard are normal, high, or low compared to what other people running AI at home are spending.

Adjacent lessons on Clawdemy

Where this sits in the track and what comes next.

• API keys and the OAuth path (lesson 3). CostGuard sits on top of the BYOK billing model that lesson established, and the Codex OAuth path is exactly why OAuth sessions show $0 against the cap in this lesson.

• How Clawless remembers (and forgets) (lesson 6). The “memories travel with messages to the provider” rule from that lesson is one specific case of the broader data-path posture this lesson covers in full. Read together for the complete privacy picture.

• The Privacy and Local-First AI track (a separate track on Clawdemy). The architectural framing in this lesson is the briefest possible version. The privacy track unpacks what each piece means, where the threats are, and how to reason about your data path across any AI tool, not just Clawless. Recommended if the data-path question is what brought you to AI cautiously in the first place.