724-office-ai-agent

Warn

Audited by Socket on Sep 14, 2026

1 alert found:

Security
SecurityMEDIUM
SKILL.md

SUSPICIOUS. The skill’s broad capabilities are mostly consistent with a self-hosted agent, but its footprint is high-risk: it executes arbitrary Python, can hot-load self-created tools, accepts webhook input, stores long-term memory, and may launch extra MCP tooling. Installation trust is weakened because ara.so publishes the skill while the actual code comes from an unrelated personal GitHub repo with no pinning or release verification. This looks more like a risky third-party agent wrapper than confirmed malware.

Confidence: 90%Severity: 72%
Audit Metadata
Analyzed At
Sep 14, 2026, 10:36 AM
Package URL
pkg:socket/skills-sh/reason-machines%2Ftrending-skills%2F724-office-ai-agent%2F@1c93d34801ff85cafa789df8096a77d62928388f211a9f121593ccb968821162