Fetch.ai has launched the Agent Execution Verification System (AEVS), a tool designed to provide a tamper-evident and independently verifiable receipt for every action executed by AI agents. AEVS addresses the issue of unreliable logs, which can be altered or misrepresented, by ensuring that each action taken—such as processing a refund—is securely logged and can be publicly verified. The system integrates seamlessly with popular frameworks like LangChain and MCP, making implementation straightforward with just two lines of code added. This innovation allows for enhanced transparency and accountability in AI operations, reinforcing compliance with regulatory standards by creating an accessible audit trail that can be verified without the need for special tools.

AEVS: AEVS is an open-source SDK that intercepts AI agent tool calls to generate cryptographically signed, hash-chained receipts, providing tamper-evident proof of every action without altering existing codebases. It supports automatic detection of frameworks like LangChain and MCP, with local buffering for reliability during network issues. Fetch.ai has just launched AEVS in beta, making it publicly available via PyPI and GitHub for developers building verifiable agent systems.
Fetch: Fetch.ai is a platform specializing in autonomous AI agents, offering tools for discovery, orchestration, and secure collaboration in an agent-based economy. It provides solutions like ASI:One for personal AI coordination and FetchCoder V2 for developing reliable multi-agent applications. In this news item, Fetch.ai announces the release of AEVS to enable auditable and verifiable executions for AI agents across engineering, compliance, and deployment teams.
Agent Execution Verification System: The Agent Execution Verification System, or AEVS, is a transparent audit tool developed by Fetch.ai that creates publicly verifiable receipts for AI agent actions using HMAC signing and end-to-end hash chains. It ensures integrity against log tampering or hallucinations by anyone, including auditors, without requiring re-execution of tools. This launch by Fetch.ai introduces the system as a simple two-line integration for existing agent frameworks.

`json
{
“Ecosystem Alignment”: “AEVS integrates with Fetch.ai’s platform and is compatible with provider-agnostic agents.”,
“Framework Compatibility”: “AEVS auto-detects and works with frameworks like LangChain and MCP, enabling seamless tool call interception.”,
“Verification Accessibility”: “Receipts can be verified independently using a public explorer with reference IDs, without needing authentication or special tools.”
}
`