Upgrades

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection risk. The skill processes untrusted data from external sources.
  • Ingestion points: YouTube transcripts and 30+ Anthropic-related web sources.
  • Boundary markers: None identified in the provided skill file.
  • Capability inventory: Execution of local scripts via bun and network requests to localhost.
  • Sanitization: No sanitization of external content is described.
  • [EXTERNAL_DOWNLOADS] (SAFE): Monitors established sources from the Anthropic ecosystem, including trusted GitHub organizations like anthropics.
  • [COMMAND_EXECUTION] (SAFE): Uses curl for local service notifications to localhost and bun to run internal helper tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:26 PM