stacktree-publish

SUSPICIOUS. The core HTML publishing flow is coherent and mostly stays on official Stacktree domains, but the skill expands into autonomous payment/provisioning actions and relies on an npx-installed toolchain from a personal publisher account. This is not confirmed malware, but it carries meaningful security and real-world action risk beyond a minimal publishing helper.

Moderate-to-high risk. This module extracts JavaScript from input .dc.html (<script data-dc-script>) and executes it via vm.runInContext, creating an “arbitrary code execution at build time” pathway if the input artifacts are not fully trusted. Additionally, template interpolation/loop rendering inserts computed values into HTML without escaping, enabling potential HTML/JS injection in the generated review page. No direct network/cryptomining/backdoor behavior is evident in this fragment, but the execution + HTML embedding combination warrants careful trust boundaries and sandbox hardening review.

SUSPICIOUS. The skill is internally aligned with its purpose, but it enables autonomous spending and live publication, and it sends registrant PII to an external registrar with limited verifiable provenance. No direct malware indicators or hidden execution appear, but the real-world action scope makes this a high-risk skill.

SUSPICIOUS. The skill's sharing behavior is broadly aligned with its stated purpose and the referenced Stacktree domains/tools look same-service and legitimate, but the actual publish mechanism depends on an unseen local script that can exfiltrate content or credentials without independent verification. The main risk is proportionate but high-impact outbound publication plus unverifiable execution details, not confirmed malware.