Code

PYTHON · GO · MOSTLY AUTHORIZATION

small tools and demos, most of them about stopping AI agents from reading things they shouldn't.

langchain-spicedb

Adds SpiceDB permission checks to LangChain and LangGraph, so your RAG pipeline only answers from documents the user can actually open.

  • Python
  • PyPI
  • Python package
  • LangChain
  • RAG
pip install langchain-spicedb
view the docs →

testcontainers-spicedb

Spins up a throwaway SpiceDB in Docker for your integration tests, with your schema already loaded. Nothing to install, nothing to clean up.

  • Python
  • PyPI
  • Testing
  • Testcontainers
  • Docker
pip install testcontainers-spicedb
view on PyPI →

steampipe-plugin-github

Query GitHub with SQL. Repos, issues, pull requests and users show up as tables you can join like any other.

  • Go
  • GitHub
  • ★ 87
  • Apache 2.0
  • SQL
steampipe plugin install github
view on GitHub →

agentic-rag-authorization

An AI agent that can only retrieve what it's explicitly allowed to see, because the check happens in code, not in the prompt. Built with LangGraph, SpiceDB and Weaviate.

  • Python
  • LangGraph
  • SpiceDB
  • Agents
  • Permissions
git clone https://github.com/authzed/examples && cd examples/agentic-rag-authorization
view on GitHub →

↑ the holes really spell each project's name, in IBM 029 punch code. you're welcome.