Skillv1.0.0

ClawScan security

Kart Io Picture Book Wizard · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 14, 2026, 1:08 PM

Verdict: Benign
Confidence: high
Model: gpt-5-mini
Summary: The skill is internally consistent with its stated purpose (an authoring/templating engine that produces bilingual picture-book text and image prompts); it requires no credentials or external installs and the included code matches the documentation, though you should review local file-write behavior and the watermark-removal prompt directive for policy/legal considerations.
Guidance: This skill appears to be what it claims: a local authoring engine that builds bilingual stories and detailed image prompts. Before installing or running: 1) review the two small shell scripts (cleanup-duplicates.sh, doc-stats.sh) and the engine Python files if you need to confirm they only write local files and don't call external services; 2) be aware the skill intentionally appends a long "no watermark/no signatures/no logos" phrase to image prompts — using prompts designed to remove watermarks or attribution can violate image provider terms or copyright rules; 3) the skill targets the 'banana nano' generator with its prompts but does not itself call banana (you would need to add your own integration and API key if you want automated image generation); and 4) if you plan to allow autonomous invocation, note it will be able to create files in the working directory. If any of these are unacceptable, inspect or sandbox the skill before use.

Review Dimensions

Purpose & Capability: okName/description (picture-book prompt + bilingual text generator optimized for 'banana nano') align with the shipped files: a Python rule engine, documentation, templates, and small helper scripts. The code implements validation, prompt assembly, CCLP markers, and local file output which are expected for this functionality.
Instruction Scope: noteSKILL.md and docs describe generating bilingual story pages, image prompts, and saving outputs to ./output/picture-books. Instructions and code only reference local files, content-safety rules, and prompt assembly. They do not request unrelated system files or secrets. Note: the skill mandates adding a long 'watermark prevention' phrase to prompts (explicit instruction to produce images with 'no watermark / no signatures / no logos'), which has policy/legal implications when used to remove existing watermarks/attribution.
Install Mechanism: okNo install spec and no external downloads; the package is self-contained. That minimizes supply-chain risk. The README shows a GitHub repo URL for manual cloning, but there's no remote install step executed by the skill itself.
Credentials: okThe skill declares no required environment variables, credentials, or config paths. It references the 'banana nano' generator as the target for its image prompts but does not perform network calls or require an API key in the provided code — the skill is an authoring tool, not an integration client, so the lack of banana API credentials is coherent.
Persistence & Privilege: okalways is false and the skill doesn't request capability to modify other skills or global agent settings. It writes outputs to a local path (./output/picture-books) which is normal for a content generator; review file-write locations if you are concerned about disk usage or data persistence.