pi-cli

Fail

Audited by Socket on Feb 27, 2026

2 alerts found:

MalwareObfuscated File
MalwareHIGH
SKILL.md

The pi-cli manifest is coherent with its stated purpose as a unified project analysis CLI. The only notable concern is the presence of explicit destructive Git commands in a CRITICAL safety section; this is acceptable as a warning but requires robust runtime safeguards to ensure it is never executed inadvertently. Absent concrete executable code or credential handling in the fragment, the overall risk is low-to-moderate (benign with safety caveats) rather than malicious.

Confidence: 95%Severity: 90%
Obfuscated FileHIGH
lib/audit/fix.js

The snippet contains no explicit malware or obvious backdoor. However, it presents a moderate-to-high supply-chain and operational risk: it sends file context to an external LLM runner and applies LLM-generated patches verbatim using unsafe string replacement without line-accurate patching, backups, or validation. If the LLM runner or its network channel is compromised, or the LLM is coerced to produce malicious changes, this tool can inject arbitrary code and leak sensitive data. Recommended mitigations: perform line-based/patch-aware updates (apply only when file slice matches exact lines), create backups and use atomic writes, validate/parse LLM JSON strictly and run syntactic checks (lint/parse/compile) on modified files, redact secrets before sending prompts, require manual approval for non-trivial changes, and harden/verify the LLM runner implementation and its network endpoints.

Confidence: 98%
Audit Metadata
Analyzed At
Feb 27, 2026, 08:52 AM
Package URL
pkg:socket/skills-sh/mkSteady%2FmkSkills%2Fpi-cli%2F@a22867474ac401d7eed748dfa37364e417ad6a0a