QWED Protocol โ Production Verification Layer for Enterprise AI & Agents
4.5K
Neurosymbolic AI Verification Layer
Verify LLM outputs with deterministic engines (SymPy, Z3, SQLGlot).
QWED is an open-source Neurosymbolic AI Verification Layer designed to make Large Language Models (LLMs) production-ready. It acts as a deterministic "Judge" that validates LLM outputs against mathematical proofs, logic constraints, SQL schemas, and code security rules before they reach your users.
SymPy.Z3 Prover.SQLGlot.docker pull qwedai/qwed-verification:latest
docker run -p 8000:8000 qwedai/qwed-verification:latest
The API will be available at http://localhost:8000.
curl -X POST "http://localhost:8000/verify" \
-H "Content-Type: application/json" \
-d '{
"engine": "math",
"query": "What is the integral of x^2?",
"llm_output": "3x"
}'
Response:
{
"verified": false,
"explanation": "Incorrect. The integral of x^2 is x^3/3.",
"correction": "x^3/3"
}
We welcome community contributions.
Ways to help:
Resources:
QWED is not just a verifier โ it is a protocol layer for AI safety.
Specialized Packages:
License: Apache 2.0โ
Maintained by: QWED AI
Content type
Image
Digest
sha256:1386bf1b4โฆ
Size
111.8 MB
Last updated
28 days ago
docker pull qwedai/qwed-verificationPulls:
150
Last week