Skip to content

Latest commit

 

History

History
82 lines (53 loc) · 1.99 KB

README.md

File metadata and controls

82 lines (53 loc) · 1.99 KB

permit-langflow

AstraDB instructions:

  1. Create a database
  2. Create Vector-enabled collection with Mistral Embeddings
  3. Leave the collection empty

Permit.io instructions:

  1. Resources with resource roles

    image

  2. Resource relations

    image

    image

    image

  3. Role derivation relations

    image

    image

    image

  4. Policy

    image

  5. Users

    image

  6. Instances

    image

LangFlow instructions:

  1. Create a new blank project

  2. Create a confidential Global Variable in Settings for Permit Key

    image

  3. 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
  4. Add Mistral API key

    1. Create a free Mistral account
    2. Get your API key from https://console.mistral.ai/api-keys/
    3. Add the key to Mistral components
  5. Run ingestion flow:

    1. Select the AstraDB database and collection
    2. Click Play button
    3. Check if the collection is populated with vectors
  6. Install Permit with pip install component

    • make sure the installation succeeds by verifing output Installation status
  7. 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
  8. Use ngrok to create a tunnel from http://localhost:7766 to an external endpoint provided by ngrok:

    ngrok http http://localhost:7766
  9. Run the chat flow by either:

    1. Opening the Playground (bottom right corner)
    2. Clicking the Play icon button on the last component
  10. Example queries:

    image

    image