Documentation / Quickstart
Quickstart Guide
From zero to a running confidential enclave in under 5 minutes. Requires a supported cloud environment (Intel TDX, AMD SEV-SNP, or AWS Nitro).
1
Install the SDK
Install the Almure Python SDK using pip. The package includes the CLI tool for local development and testing.
shell
$ pip install almure-sdk
Collecting almure-sdk
Downloading almure_sdk-0.9.4-py3-none-any.whl (48 kB)
Successfully installed almure-sdk-0.9.42
Write your enclave config
Create an enclave.yaml file in your project root. This defines the workload image, memory allocation, and attestation policy.
enclave.yaml
enclave_image: ./inference.tar
memory_mb: 2048
attestation_policy:
sealing: measurement-bound
min_isvsvn: 3
output_encryption: true3
Launch the enclave
Use the SDK to launch your workload and retrieve the execution result. The SDK returns an EnclaveResult object containing the output and the attestation report.
launch.py
almure
pathlib Path
client = almure.Client(api_key=os.environ["ALMURE_API_KEY"])
result = client.launch(
config=Path("enclave.yaml"),
input_data=dict(records=encrypted_batch)
)
print(result.output) # decrypted inference result
print(result.attestation.mrenclave) # hex measurement hash4
Verify the attestation
Compare the returned mrenclave against your known-good measurement to confirm the correct workload ran. This is the step your compliance team reviews.
verify.py
APPROVED_MRENCLAVE = "3a8f2c1d9e4b7a06f5d38c12e91b4a7f..."
result.attestation.mrenclave == APPROVED_MRENCLAVE, "Measurement mismatch"
print("Attestation verified — workload matched approved build")Next steps
- Browse the full API reference for all endpoints
- Understand attestation and how to use the report in compliance reviews
- Request a pilot scoping call