Stop Burning Tokens — OpenClaw Cost Diagnostic Primer

Diagnostic framework to identify and reduce token waste in OpenClaw deployments, cutting costs by applying efficient data handling and model use practices.

Security Scan

Instruction-only primer about reducing token waste; it contains best-practice recommendations and makes no requests for credentials, binaries, or installs, and therefore appears internally consistent with its stated purpose.

Purpose & Capability

The name and description (token-cost diagnostics) match the SKILL.md content. The skill is purely advisory and requests no env vars, binaries, or install steps that would be unrelated to cost optimization.

Instruction Scope

SKILL.md contains diagnostic questions, cost comparisons, and operational recommendations (indexing, caching, model routing). It does not instruct the agent to read arbitrary files, exfiltrate data, or access environment variables or system paths beyond its advisory scope.

Install Mechanism

No install spec and no code files. Being instruction-only means nothing is written to disk or executed by the skill itself.

Credentials

The skill declares no required environment variables, credentials, or config paths. Recommendations (e.g., cache stable context) are conceptual and do not require extra secrets.

Persistence & Privilege

Flags show always:false and normal invocation behavior. The skill does not request persistent presence or modify other skills or system configuration.

Assessment

This skill is advisory only and appears safe to read/use: it gives best-practice diagnostics for reducing token costs and does not request credentials or install code. Before applying recommendations, (1) verify any external links or commercial offerings independently (the SKILL.md links to a ShopClawMart listing), (2) test changes (model routing, session caps, caching) in a staging environment to confirm actual savings, (3) be mindful of privacy/retention when caching or indexing context — avoid storing sensitive PII unless your storage and retention policies are reviewed, and (4) instrument and measure token usage before/after changes so you can validate claimed cost deltas. If you plan to operationalize the recommendations (implement caching, connector audits, or automated routing), review those implementation details for data-handling and credential use — the primer itself does not perform those actions.

SKILL: Token Cost Intelligence — Free Primer

Source: Production agent stack running at $0.91/day (down from $8–10/session) Domain: Token cost optimization, OpenClaw deployments Type: Free primer

---

THE CORE TRUTH

The models are not expensive. Your habits are.

Most OpenClaw operators are spending 8–10x more than they need to. This primer gives you the diagnostic framework to find out where you're leaking.

---

THE "STUPID BUTTON" — 6 DIAGNOSTIC QUESTIONS

Run these before every session:

1. Are you feeding raw PDFs/images when you only need text? Screenshots are the worst offender. Copy-paste or convert to Markdown. A 4,500-word PDF = 100,000+ tokens raw. The same content in Markdown = 4,000–6,000 tokens. ~20x reduction.

2. When did you last start a fresh conversation? Every new turn re-sends the entire conversation history. 30-turn threads don't just feel inefficient — they are. 10–15 turn cap, then summarize and start fresh.

3. Are you using the most expensive model for everything? Opus for formatting and proofreading is a Ferrari to the grocery store. Haiku handles light tasks at 1/30th the cost.

4. Do you know what's loading in context before you type? Each loaded plugin = silent token tax per session. Documented case: 50,000 tokens consumed before the first keystroke. Audit your connectors. Disable what you don't use.

5. Are you caching stable context? (API builders) Cache hits on Opus: $0.50/M vs $5.00/M standard = 90% discount. System prompts, tool definitions, persona instructions → all cacheable. If you're not caching, you're paying full price for the same tokens every call.

6. How are you handling web search? Native model web search is token-heavy. MCP-routed alternatives return structured results at a fraction of the cost. Know what you're paying per search.

---

COST COMPARISON (CONCRETE)

Session Type	Input Tokens	Output Tokens	Cost (Opus pricing)
Sloppy (raw PDFs, 30-turn sprawl, Opus-everything)	800K–1M	150K–200K	$8–$10
Clean (markdown, 10-turn cap, tiered models)	100K–150K	50K–80K	~$1
Reduction	~8x	~3x	8–10x

Scaled to a team of 10 for one month:

• Sloppy habits: ~$2,000/month
• Clean habits: ~$250/month
• Same output volume.

---

5 AGENT COMMANDMENTS

For anyone running OpenClaw agents at any scale:

1. Index your references. Agents get relevant chunks, not raw document dumps. Dumping full documents per agent call is architectural waste.

2. Pre-process context before it hits the window. Chunk, summarize, and clean before ingestion. If the model's first tokens are spent parsing your bad preprocessing, you failed.

3. Cache your stable context. System prompts, tool definitions, persona instructions, reference material → all cacheable. Thousands of agent calls per day without caching is pouring money out.

4. Scope each agent to minimum viable context. Planning agent doesn't need the full codebase. Editing agent doesn't need the project roadmap. Passing everything to every agent is measurable waste — and models perform worse drowning in irrelevant context.

5. Measure what you burn. Instrument all agent calls: input tokens, output tokens, model mix, cost ratio. You cannot optimize what you don't measure.

---

Full framework with anti-patterns by tier, tiered model routing, and confirmed production delta available in Token Cost Intelligence on Claw Mart.