subject-learning-assistant

SUSPICIOUS: the skill’s purpose is coherent, and there is no clear credential harvesting or overt exfiltration in the instructions. However, its core functionality depends on an ambiguously sourced external CLI (memocli/memories-off) whose publisher and official install path cannot be cleanly verified from the evidence, making the skill high risk from a supply-chain trust perspective.

The code is a local knowledge-base dashboard server. It does not show malicious behavior or obfuscation. It does present a meaningful information-disclosure risk because it binds to all interfaces, exposes parsed knowledge-base data without authentication, enables unrestricted cross-origin reads, and delegates other paths to a static file server. Restrict binding to localhost or add authentication and tighter routing before using it on an untrusted network.

The code appears to implement a live D3 dashboard and contains no clear malicious behavior. It has a concrete DOM XSS risk because server-provided fields are inserted into innerHTML and tooltip.html() without escaping or sanitization. Use textContent for plain text or a trusted HTML sanitizer with strict content controls. The polling and visualization behavior are otherwise expected.