asking-with-evidence

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill uses Bash to invoke the watch-skill utility with user-controlled arguments, a pattern typically associated with command execution risks.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input which is passed to the command-line interface. Ingestion points: user questions and search phrases in SKILL.md. Boundary markers: documentation suggests double-quoting user inputs. Capability inventory: Bash and Read tools. Sanitization: no explicit sanitization logic is provided.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 12:04 AM
Security Audit — agent-trust-hub — asking-with-evidence