fabio-rovai

fabio-rovai / tardygrada

Public

A programming language where every agent output is cryptographically verified. 194KB binary. Zero dependencies. Pure C11. Coq-proven BFT. tardy terraform replaces LangChain/CrewAI/AutoGen in ~15 lines.

agent-framework ai-agents ai-safety byzantine-fault-tolerance c
10
0
100% credibility
Found Apr 04, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
C
AI Summary

Tardygrada is a compact programming language and runtime that compiles programs into verifiable AI agents using cryptographic proofs, knowledge grounding, and consensus to ensure factual accuracy.

How It Works

1
🔍 Discover Tardygrada

You hear about a super-small way to build trustworthy AI helpers that always check facts before believing them.

2
⚡ Get it ready fast

Download and start it up in under a second—no complicated setups needed.

3
✅ Test a fact

Ask if something real like 'Doctor Who started at BBC' is true, and it double-checks against trusted knowledge.

4
✨ Shrink big projects

Turn a massive AI toolbox into a tiny, safe version that proves everything it says.

5
🤖 Build your helper

Write a simple note describing your smart assistant, like a medical advisor that verifies advice.

6
🌐 Share it safely

Put your helper online so others can talk to it securely.

🎉 Truthful AI magic

You now have reliable agents that never guess or lie—they prove every claim with facts.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is tardygrada?

Tardygrada is a tiny programming language for AI agents where every output claim gets cryptographically verified: decomposed to triples, grounded in ontologies, and BFT-consensus checked across 8 layers before becoming an immutable Fact. Built in pure C11 as a 194KB zero-dependency binary, it compiles .tardy programs to MCP servers via simple CLI commands like `tardy run "claim text"` or `tardy serve medical.tardy`. The `tardy terraform` tool shrinks LangChain/CrewAI/AutoGen repos to 15 lines.

Why is it gaining traction?

It crushes Python agent frameworks on size (212KB vs 150K+ lines) and trust: Coq-proven BFT stops hallucinations/laziness with real ed25519 signatures and mprotect immutability, delivering 0% false positives offline. Devs dig the speed—5M verified reads/sec—and terraform magic for github programming ai prototypes, plus honest "unknown" fallbacks in this github programming language.

Who should use this?

AI engineers building verifiable multi-agent workflows, like medical diagnosis chains or fact-checking bots, who need grounded claims without LLM drift. Teams evaluating github programming projects for production agents, or backend devs swapping bloated LangChain for lean C binaries in edge deployments.

Verdict

Grab it for agent verification experiments—the 194KB demo flies, but with 10 stars and 1.0% credibility, it's raw; run `make && tardy run` tests first before terraforming real repos. Solid proofs beat hype.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.