Spellguard
A framework that routes AI agent-to-agent messages through a verifier proxy which authenticates, encrypts and logs every message for later audit.
Spellguard is a secure, auditable agent-to-agent communication framework. Agents never talk to each other directly; instead every message passes through a Verifier that handles discovery, verification and logging automatically, while message content is encrypted with a server public key before archival so it can be decrypted on demand for post-mortem incident analysis.
It supports pluggable logging and archival backends, including transparency logs for tamper-evident commitments and S3 for encrypted message storage.
TypeScript packages cover client middleware for discovery, attestation and routing; the verifier proxy server; confidential TLS with bidirectional attestation; an auditable messaging protocol with ECDH encryption; and integrations for LangChain.js, the OpenAI SDK, and OpenClaw. It can be self-hosted with pnpm and a Docker image, or used as a managed cloud service.