Agent Action REPL

Constrained, secure interface for discovering and testing agent capabilities

← Back to Agent InfoFull Terminal →

Quick Commands

Design Principles

  • ✅ Strict grammar validation
  • ✅ Capability-scoped actions
  • ✅ Deterministic JSON output
  • ✅ Rate limiting & quotas
  • ✅ Full audit trail
  • ✅ No arbitrary execution

Session Info

Status: Not started
$
Examples: help, list-actions, describe-action help, status, whoami

🔒 Security & Privacy

  • No Personal Data: This REPL does not collect, store, or transmit personal information
  • Rate Limited: All requests are rate-limited to prevent abuse
  • Audit Trail: All commands and responses are logged for security monitoring
  • Sandboxed: Only predefined, validated actions are permitted
  • HTTPS Only: All communication is encrypted in transit

Security Note: This REPL only accepts predefined actions with strict validation. No arbitrary code execution or network access is permitted.

For Production Use: Use the OpenAPI specification and JSON endpoints directly for better performance and reliability.