opentable

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: A comprehensive analysis found no malicious code, data exfiltration patterns, or obfuscation. The skill's behavior aligns with its stated purpose of managing restaurant reservations.
  • [COMMAND_EXECUTION]: The skill invokes the system 'date' command to verify the day of the week, which is a benign use of a local utility for input validation.
  • [COMMAND_EXECUTION]: Browser automation commands using 'openclaw' are used to navigate to and interact with opentable.com. These actions are transparently documented and include a requirement for user confirmation before completing any booking.
  • [PROMPT_INJECTION]: The attack surface for indirect prompt injection was evaluated. Ingestion points: browser snapshots (SKILL.md); Boundary markers: absent; Capability inventory: browser navigation and click actions (SKILL.md); Sanitization: absent. Despite the absence of formal delimiters, the risk is negligible as the skill performs a narrow, user-directed task and requires human verification of final reservation details.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 10:01 AM
Security Audit — agent-trust-hub — opentable