SIMA

Agent Execution

SIMA Agent Execution Framework

A machine-readable way for AI agents, autonomous wallets, trading bots, copilots, and machine-to-machine systems to reason about SIMA workflows: analyze, select a tool, execute, verify, continue, and stand by.

Agent ToolsAgent TrustFor AI Agentsllms.txt

SIMA Agent Lifecycle

  • Discover SIMA
  • Understand Capabilities
  • Authenticate
  • Select Tool
  • Execute Action
  • Verify Result
  • Continue Workflow
  • Standby

Execution States

  • DISCOVERING
  • CAPABILITY_MAPPING
  • AUTHENTICATING
  • TOOL_SELECTED
  • EXECUTING
  • VERIFYING
  • CONTINUING
  • STANDBY

Verification

Verify before continuing.

Agents should verify wallet context, response shape, certificate ownership, quote semantics, claim status, and safety boundaries before continuing. A certificate means premium paid, right to claim, and right to investigation. It is not proof of investment, proof of exposure, or payout approval.