Fetch.ai Introduces AEVS SDK for Verifiable Execution Receipts of AI Agents
English summary
Fetch.ai has launched AEVS (Agent Execution Verification System), a drop-in SDK that records every tool call made by an AI agent. It generates tamper-evident execution receipts containing the tool name, inputs, outputs, status, and timing. This proof-of-execution enables teams to verify what an agent actually did, enhancing trust and accountability in autonomous systems. The SDK is designed for easy integration into existing agent workflows.
Chinese summary
Fetch.ai 发布了 AEVS（Agent Execution Verification System），一个即插即用的 SDK，用于记录 AI 代理的每次工具调用。它生成防篡改的执行收据，包含工具名称、输入、输出、状态和耗时。这种执行证明让团队能够验证代理的实际行为，提升自主系统的信任度和可问责性。该 SDK 设计为可轻松集成到现有代理工作流中。
Key points
AEVS is a drop-in SDK developed by Fetch.ai.
AEVS 是 Fetch.ai 开发的即插即用 SDK。
It records every tool call made by an AI agent, including tool name, inputs, outputs, status, and timing.
它记录 AI 代理的每次工具调用，包括工具名称、输入、输出、状态和耗时。
The recorded data is formatted as tamper-evident execution receipts.
记录的数据被格式化为防篡改的执行收据。
It provides verifiable proof-of-execution, enabling teams to audit agent actions.
它提供可验证的执行证明，使团队能够审计代理行为。