claw-token-cost-analyzer

Analyzes Claw AI workflows to estimate token usage, cost, detect runaway loops, and suggest optimizations to prevent unexpected API expenses.

Security Scan

The skill is an instruction-only analyzer that asks the user to paste workflows/prompts and contains no code, no installs, and requests no credentials — its declared purpose aligns with its instructions.

Purpose & Capability

Name and description match the content of SKILL.md: it analyzes Claw workflows for token usage, runaway loops, and suggests optimizations. No unrelated capabilities, binaries, or credentials are requested.

Instruction Scope

SKILL.md limits itself to asking the user to paste a workflow/prompt and describing checks and expected output format. It does not instruct the agent to read files, access environment variables, call external endpoints, or perform system-level actions.

Install Mechanism

No install spec and no code files — instruction-only. This is the lowest-risk install posture and nothing is written to disk or downloaded.

Credentials

No environment variables, credentials, or config paths are required. The SKILL.md does not reference any secrets or external service keys.

Persistence & Privilege

always is false and the skill is user-invocable. It does not request persistent system presence or modifications to other skills/configuration.

Assessment

This skill is instruction-only and coherent with its stated purpose — it asks you to paste workflows/prompts and returns an analysis. It does not request credentials or install anything. Before relying on its output, test with known examples and compare its estimates to your actual billing data (estimations can be imprecise). Additionally, because it is just an instruction template (no code), ensure the agent runtime you use executes the analysis as you expect and does not add any external calls; review results before taking automated remediation steps. The license is MIT-0 (no attribution).

Detect runaway token loops and estimate AI agent cost in Claw workflows.

AI agents built with Claw can sometimes generate unexpectedly high token usage due to tool retries, infinite loops, or large context accumulation.

This skill analyzes a workflow or prompt and estimates:

• Token usage
• Potential cost per run
• Runaway loop risk
• Optimization suggestions

It is designed for AI agent builders who want to prevent surprise API bills before deploying their automation.

---

Example input

Agent researches competitor pricing and retries web search until confident.

---

How to use

Paste your Claw workflow, task description, or prompt.

Example:

Agent researches competitors and repeatedly calls a web search tool until pricing data is found.

---

What the analyzer checks

The analyzer looks for common patterns that lead to high token usage:

• Infinite loops
• Tool retry chains
• Recursive agent calls
• Large context growth
• Expensive model usage

---

Output format

Output must follow this format:

AI Agent Cost Analysis

Risk level: (Low / Medium / High)

Estimated tokens per run:

Estimated cost per run:

Issues detected:

Optimization suggestions:

---

Example Output

AI Agent Cost Analysis

Risk level: Medium

Estimated token usage: 5k – 20k tokens per run

Estimated cost: $0.10 – $0.60 per run

Potential issues detected:

• Possible tool retry loop
• Large context accumulation

Optimization suggestions:

• Limit tool retries
• Use smaller models for research tasks
• Add step limits to prevent runaway loops

---

Why this matters

AI agents can accidentally generate runaway token usage when tools retry repeatedly or workflows contain recursive steps.

Even a small mistake can multiply costs quickly.

Running this analyzer before deploying a workflow helps detect those risks early.

---

Tip

If you run AI agents in production, consider using a gateway that can enforce budget limits and block abnormal calls automatically.

Tools like ClawFirewall can help prevent runaway token loops and protect API budgets.

---

License

MIT-0 (No Attribution)