qmd
Warn
Audited by Socket on Mar 18, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
The skill fragment is largely coherent with its stated purpose of local Markdown search and optional semantic search. However, the install path (direct bun install from a GitHub URL) and the implied auto-download of local GGUF models introduce supply-chain and integrity risks that should not be dismissed. Without explicit verification of model integrity (checksums, signed artifacts) and without using a trusted registry for installation, this pattern elevates risk. Overall, the footprint is suspicious rather than clearly malicious: benign functionality is present, but the distribution and model acquisition methods could be exploited if the sources are compromised or misconfigured.
Confidence: 75%Severity: 75%
Audit Metadata