sync-global-rules
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local shell scripts (
check_updates.shandsync.sh) to perform version checks, create file backups, and synchronize configuration data. - [EXTERNAL_DOWNLOADS]: Fetches markdown files and repository metadata from the author's GitHub repository (
nangongwentian-fe/Awesome-GlobalRule) using the GitHub CLI (gh api). These downloads are handled neutrally as they originate from the vendor's own established repository. - [INDIRECT_PROMPT_INJECTION]: This skill downloads and installs remote markdown files that are subsequently used as 'global rules' for the AI agent (
CLAUDE.md,AGENTS.md). This creates an attack surface where the agent's behavior could be altered by external content without direct user verification of the rule logic. - Ingestion points: Remote markdown files from the
Awesome-GlobalRulerepository. - Boundary markers: Absent; the downloaded files are installed directly as system-level instructions.
- Capability inventory: The skill scripts have the ability to write to sensitive local tool directories (
~/.claude/and~/.codex/) and create local backups. - Sanitization: No validation or sanitization of the downloaded markdown content is performed before it is applied to the agent's active configuration.
Audit Metadata