tardygrada
Trust infrastructure for AI agents. Know who produced a value, when, and that it hasn't been tampered with. Zero dependencies. Pure C.
Tardygrada is a programming language that compiles to MCP servers, where every value carries cryptographic proof of its creator, creation time, and integrity. It guarantees integrity via mprotect, SHA-256, ed25519, and BFT consensus (proven in Coq), provenance with signed metadata, and a full audit trail. However, it does not guarantee truth; it only ensures the value has not been changed, providing 100% proof even after 100 handoffs among agents.
Features
Compatibility
Quick start
Use cases
Alternatives
Related searches
Comments
- QQuinn WhiteMay 11, 2026
Trust infrastructure for AI agents — provenance, timestamp, and integrity verification.
- Kendall LeeMay 7, 2026
Works as a trust layer that other agent frameworks can build on.
- EEllis JohnsonApr 11, 2026
Good for systems where knowing who produced a value and when is a requirement.
- OOakley RiveraMar 25, 2026
Provenance tracking builds audit trails for AI-generated outputs.