AEVS

AEVS is a developer tool that adds a layer of verifiable proof to AI agent operations. Its core function is to generate a signed, tamper-evident receipt for every tool an agent calls—like making a payment, updating a database, or sending an email. This creates a clear, cryptographic record of what the agent did, what inputs it used, and what the result was, moving beyond just the agent's text response.

It works as a drop-in Python SDK that integrates with popular agent frameworks like LangChain and MCP. Once enabled, it automatically intercepts tool calls, captures the details, and seals each event with an ECDSA P-256 signature. These receipts are linked in a hash chain, so any alteration or missing step is immediately apparent. You can verify execution independently via a simple public API call or a web-based explorer, without needing to re-run the agent or share your codebase.

This tool is built for developers and teams who need to ship reliable, auditable AI agents, especially in production environments. It's particularly useful for scenarios where proving an action happened is critical, such as in financial services, healthcare, or any compliance-heavy workflow. By providing configurable data visibility—from fully public receipts to proof-only modes that never store sensitive payloads—AEVS gives teams the flexibility to meet both transparency and security requirements.