skill-installer

Warn

Audited by Socket on Sep 15, 2026

2 alerts found:

Securityx2
SecurityMEDIUM
SKILL.md

SUSPICIOUS: the stated purpose matches installing skills, but the skill is inherently higher risk because it installs other skills into trusted agent directories. The main concerns are transitive trust and unclear catalog provenance, plus mismatches with official Anthropic/OpenCode install documentation. No direct credential theft or exfiltration is evident in the provided content.

Confidence: 91%Severity: 74%
SecurityMEDIUM
scripts/install_skill.py

The code is primarily a skill installation utility and contains no clear direct malware such as reverse shells, credential harvesting, cryptomining, or covert exfiltration. However, it has significant path validation weaknesses. A malicious catalog response can potentially write files outside the temporary destination, and crafted local skill names or uninstall arguments can escape the intended harness directory and overwrite or delete arbitrary paths. These issues should be fixed by resolving paths and enforcing containment, rejecting absolute paths and traversal components, and validating skill names before installation or removal. The code fragment also appears incomplete and may not run as provided.

Confidence: 96%Severity: 72%
Audit Metadata
Analyzed At
Sep 15, 2026, 04:16 PM
Package URL
pkg:socket/skills-sh/skillscatalog%2Fregistry%2Fskill-installer%2F@bcd8e3d9f503bdaf23cfe6abf8d0064bd76729d91fe6234428f9f234038335fa