- Create a database
- Create Vector-enabled collection with Mistral Embeddings
- Leave the collection empty
-
Create a new blank project
-
Create a confidential Global Variable in Settings for Permit Key
-
Import
flows.json
into LangFlow- you should be ok with importing this JSON only
- other
*.py
files are added for reference as they are custom components implemented for this project
-
Add Mistral API key
- Create a free Mistral account
- Get your API key from https://console.mistral.ai/api-keys/
- Add the key to Mistral components
-
Run ingestion flow:
- Select the AstraDB database and collection
- Click Play button
- Check if the collection is populated with vectors
-
Install Permit with
pip install
component- make sure the installation succeeds by verifing output
Installation status
- make sure the installation succeeds by verifing output
-
Run the Permit Policy Decision Point (PDP) locally:
docker run -p 7766:7000 --env PDP_API_KEY=<YOUR_KEY> --env PDP_DEBUG=true permitio/pdp-v2:latest
-
Use
ngrok
to create a tunnel fromhttp://localhost:7766
to an external endpoint provided byngrok
:ngrok http http://localhost:7766
-
Run the chat flow by either:
- Opening the Playground (bottom right corner)
- Clicking the Play icon button on the last component
-
Example queries: