harari-perspective

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of markdown-based instructions and research datasets (01-writings.md through 06-timeline.md). No executable binaries, scripts, or persistent system modifications are included.
  • [PROMPT_INJECTION]: The skill implements a specific persona but includes robust constraints ('绝不假装的事') that prevent the agent from making false authoritative claims or bypassing safety filters. The 'Agentic Protocol' provides a structured reasoning flow that mitigates simple instruction-override attempts.
  • [DATA_EXFILTRATION]: No patterns of sensitive data access (e.g., credentials, SSH keys) or external network transmissions were identified. All URLs point to reputable informational or academic platforms.
  • [EXTERNAL_DOWNLOADS]: The README specifies a platform-standard installation command ('npx skills add jazzqi/harari-skill') targeting the author's official repository. This is consistent with legitimate skill distribution.
  • [COMMAND_EXECUTION]: There are no instances of shell command execution, subprocess spawning, or dynamic code evaluation in the analyzed files.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 11:15 AM
Security Audit — agent-trust-hub — harari-perspective