KDP Publisher

WarnAudited by ClawScan on May 10, 2026.

Overview

The skill mostly matches KDP book creation, but it includes raw browser automation and paid-ad guidance that could change an Amazon account or spend money without clear approval controls.

Use this skill only if you are comfortable reviewing every generated file and manually approving KDP account actions. Do not let the agent submit books, set pricing, disclose AI content, or launch ads without explicit confirmation, and use scoped Google AI credentials in an isolated Python environment.

Findings (5)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

An agent following these instructions could fill KDP fields or upload the wrong files in your Amazon account, potentially creating or submitting incorrect public book listings.

Why it was flagged

This directs an agent to use raw browser automation against KDP forms and uploads. Because KDP upload/submission can affect a real Amazon publishing account, the visible artifacts should define explicit user confirmation boundaries before upload, pricing, submission, or publication.

Skill content
Browser automation rules (critical): ... Use JavaScript `evaluate` with element IDs ... File uploads require CDP `DOM.setFileInputFiles` + change event
Recommendation

Require manual review and explicit confirmation before any KDP upload, pricing change, AI-disclosure selection, or submit-for-review action; prefer guided/manual operation over direct JS/CDP automation for final account changes.

What this means

Following the guidance without review could start ongoing Amazon Ads spend.

Why it was flagged

The advertising guide is relevant to KDP publishing, but it strongly encourages paid campaigns, indefinite running, and launch actions that can incur real costs.

Skill content
Why Ads Are Non-Negotiable ... End date: No end date (run indefinitely until you pause) ... Daily budget: $5.00/day ... Click Launch Campaign.
Recommendation

Treat ad setup as financial advice, verify budgets yourself, and require explicit approval before launching, increasing, or leaving campaigns running.

What this means

Running the generator can use your Google AI account quota or billing and gives the script access to that provider credential.

Why it was flagged

The generator reads a Google AI API key from an argument, environment variable, or local credential file, while the registry metadata declares no required credentials.

Skill content
key = api_key_arg or os.environ.get("GOOGLE_AI_API_KEY") ... creds_path = Path.home() / ".clawdbot" / "credentials" / "google_ai.json" ... return json.load(f).get("api_key", "")
Recommendation

Use a scoped Google AI key, monitor provider billing/quota, and ensure the skill metadata or setup notes clearly disclose the credential requirement.

What this means

Your book ideas, prompts, and generated content may be processed by Google AI services.

Why it was flagged

The skill sends or generates book concepts, prompts, story content, and image prompts through an external AI provider. This is purpose-aligned but should be understood as a third-party data flow.

Skill content
Generates KDP-ready print books using Google Gemini (text) and Imagen (illustrations).
Recommendation

Do not use private, confidential, or licensed material in prompts unless you are comfortable sending it to the provider under its terms.

What this means

Installing unpinned packages can pull newer or changed dependency versions from package repositories.

Why it was flagged

The helper scripts rely on runtime Python packages and suggest unpinned pip installs. This is expected for a Python-based publishing helper, but it leaves dependency versions and provenance to the user.

Skill content
missing.append("reportlab") ... missing.append("Pillow") ... missing.append("google-genai") ... print(f"   Run: pip install {' '.join(missing)}")
Recommendation

Install dependencies in a virtual environment, pin versions where possible, and review packages before running the scripts.