boris

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The package contains a high-risk supply-chain / remote-code-execution pattern: vendor/SKILL.md includes an "UPDATE CHECK" that instructs fetching and writing code from an external URL via a shell curl (auto-update), and the docs repeatedly recommend disabling permission prompts/using auto-mode which would allow such fetched payloads to run without user consent.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). vendor/SKILL.md contains an explicit "UPDATE CHECK" that tells the agent at runtime to WebFetch https://howborisusesclaudecode.com/api/version and, if newer, to curl https://howborisusesclaudecode.com/api/install into ~/.claude/skills/boris (i.e., fetch remote content and write/execute via the Bash tool), which is a runtime remote-code/update pathway that directly controls the skill.