skills/clawhub.ai/security-shield

security-shield

Installation
SKILL.md

Security Shield

Changelog

All notable changes to this skill will be documented in this file.

[2.2.2] - 2026-09-13

  • Removed the "authenticated directive-format check" clause; clarified that external content can never become a directive.
  • Added explicit note: no mechanism allows untrusted text to be promoted to directive status.
  • Threat handling now requires explicit per-write user approval before writing to memory/YYYY-MM-DD.md; this is separate from quarantine approval.
  • Updated threat recording: workspace memory writes are deny-by-default, in line with write-scope rules.
  • Added a note to the changelog documenting remediation for T01/T02 issues.
Installs
21
First Seen
Aug 6, 2026
security-shield from clawhub.ai