This repository serves as a template for creating a proof of contribution tasks using Python. It is executed on Vana's Satya Network, a group of highly confidential and secure compute nodes that can validate data without revealing its contents to the node operator.
This template provides a basic structure for building proof tasks that:
- Read input files from the
/input
directory. - Process the data securely, running any necessary validations to prove the data authentic, unique, high quality, etc.
- Write proof results to the
/output/results.json
file in the following format:
{
"dlp_id": 1234, // DLP ID is found in the Root Network contract after the DLP is registered
"valid": false, // A single boolean to summarize if the file is considered valid in this DLP
"score": 0.7614457831325301, // A score between 0 and 1 for the file, used to determine how valuable the file is. This can be an aggregation of the individual scores below.
"authenticity": 1.0, // A score between 0 and 1 to rate if the file has been tampered with
"ownership": 1.0, // A score between 0 and 1 to verify the ownership of the file
"quality": 0.6024096385542169, // A score between 0 and 1 to show the quality of the file
"uniqueness": 0, // A score between 0 and 1 to show unique the file is, compared to others in the DLP
"attributes": { // Custom attributes that can be added to the proof to provide extra context about the encrypted file
"total_score": 0.5,
"score_threshold": 0.83,
"email_verified": true
}
}
The project is designed to work with Intel TDX (Trust Domain Extensions), providing hardware-level isolation and security guarantees for confidential computing workloads.
my_proof/
: Contains the main proof logicproof.py
: Implements the proof generation logic__main__.py
: Entry point for the proof executionmodels/
: Data models for the proof system
demo/
: Contains sample input and output for testingDockerfile
: Defines the container image for the proof taskrequirements.txt
: Python package dependencies
To use this template:
- Fork this repository
- Modify the
my_proof/proof.py
file to implement your specific proof logic - Update the project dependencies in
requirements.txt
if needed - Commit your changes and push to your repository
The main proof logic is implemented in my_proof/proof.py
. To customize it, update the Proof.generate()
function to change how input files are processed.
The proof can be configured using environment variables:
USER_EMAIL
: The email address of the data contributor, to verify data ownership
If you want to use a language other than Python, you can modify the Dockerfile to install the necessary dependencies and build the proof task in the desired language.
To run the proof locally for testing, you can use Docker:
docker build -t my-proof .
docker run \
--rm \
--volume $(pwd)/input:/input \
--volume $(pwd)/output:/output \
--env [email protected] \
my-proof
Intel TDX (Trust Domain Extensions) provides hardware-based memory encryption and integrity protection for virtual machines. To run this container in a TDX-enabled environment, follow your infrastructure provider's specific instructions for deploying confidential containers.
Common volume mounts and environment variables:
docker run \
--rm \
--volume /path/to/input:/input \
--volume /path/to/output:/output \
--env [email protected] \
my-proof
Remember to populate the /input
directory with the files you want to process.
This template leverages several security features:
- Hardware-based Isolation: The proof runs inside a TDX-protected environment, isolating it from the rest of the system
- Input/Output Isolation: Input and output directories are mounted separately, ensuring clear data flow boundaries
- Minimal Container: Uses a minimal Python base image to reduce attack surface
Feel free to modify any part of this template to fit your specific needs. The goal is to provide a starting point that can be easily adapted to various proof tasks.
If you have suggestions for improving this template, please open an issue or submit a pull request.