amazon-kdp

SUSPICIOUS: the skill is broadly coherent with Amazon KDP automation and uses expected tools, but it carries medium risk because it stores reusable browser sessions and grants the agent powerful real-world publishing/account actions. No clear third-party credential harvesting or malware indicators are present in the provided material.

SUSPICIOUS. The stated purpose is coherent, but the skill routes all operations through an undocumented local API/npm script rather than verifiable Amazon endpoints or clearly identified tooling. Scope is roughly proportional to KDP metadata editing, yet install/execution trust and credential/data-flow transparency are insufficient.

SUSPICIOUS. The skill’s actions are largely aligned with KDP publishing, and the dry-run/live safeguards are proportionate, but execution trust and data-flow integrity are only partially verifiable because all meaningful behavior is hidden behind opaque local npm scripts and local /api/kdp/* endpoints rather than a disclosed official Amazon integration.

SUSPICIOUS: the stated purpose is coherent, but the skill routes sensitive account-changing actions through an undocumented local intermediary whose code, provenance, and credential/data flows are not disclosed. Medium risk from opaque execution trust rather than clear malware behavior.

No clear evidence of overt malware exists in the provided fragment. The main supply-chain/security risk is the eval-equivalent behavior: arbitrary JavaScript contained in a locally read file (fillBookDetails.js) is executed in the browser context via page.evaluate. If that file could be modified (compromised repo, build pipeline tampering, or local attacker), this becomes a high-impact execution path. Additionally, the script logs titleId and ISBN, which may be sensitive in certain environments. Review and harden the trust and integrity of fillBookDetails.js (and consider reducing/avoiding dynamic code execution).