Skip to main content

AURORA Agent Evidence SDK

PyPI License

aurora-agent is a local-first Python SDK for instrumenting declared consequential machine actions and transmitting compositional execution evidence to AURORA.

python -m pip install --upgrade aurora-agent

It provides seven compatible surfaces:

  1. Local action evidence — proposal, authorization/policy, precommit, execution, outcome, deterministic bundle export, and offline verification.
  2. AURORA ingestion — incremental runtime events, a durable SQLite outbox, idempotent transport, and finalization into an immutable AURORA Compositional Evidence graph.
  3. AI Output evidence — first-class AuroraSeal v3 records with digest-only, redacted, or full-payload capture, sealing, verification, bundle export, and signed links to sealed AI decision records.
  4. AI Decision evidence — first-class decision records that preserve operator-declared score interpretation, policy context, evidence gaps, uncertainty, sealing, public metadata verification, and offline bundles.
  5. Human Approval evidence — policy-bound approval requirements, immutable review events, reviewer eligibility, deterministic single- and multi-party process projections, and human-bound approval operations.
  6. Amendment lifecycle evidence — immutable corrections, amendments, supersessions, reversals, and withdrawals with full-successor preparation, current-head coordination, lifecycle projection, and verification bundles.
  7. Data Lifecycle evidence — retention, redaction, disclosure, tombstone, content-artifact, operation-intent, projection, and independently verifiable lifecycle-bundle surfaces.

Core properties

  • deterministic JSON canonicalization and SHA-256 commitments;
  • raw-payload minimization with DIGEST_ONLY as the default;
  • explicit REDACTED and FULL_PAYLOAD modes;
  • API keys held only in memory and never written to the outbox;
  • ordered, crash-recoverable delivery with stable idempotency keys;
  • generic Python capture surface;
  • framework-light Claude Agent SDK adapter;
  • direct observer integration with the JAKROW D3 execute_operation path;
  • final AURORA graph linked to an existing sealed AURORA record.

Self-serve quickstart

After creating a Sandbox API key at https://auroraseal.com/app/quickstart:

python -m pip install --upgrade aurora-agent
export AURORA_API_KEY="<shown-once-key>"
aurora-agent quickstart

The command creates a controlled sample record, commits evidence locally before a local SQLite ledger operation, finalizes the graph, verifies it server-side, and prints the authenticated Viewer URL.

It does not perform an external payment, message, purchase, or customer operation. The default capture mode is DIGEST_ONLY; the API key is not persisted in the outbox.

Recovery commands:

aurora-agent outbox status --outbox PATH
aurora-agent outbox flush --outbox PATH

Package identity

distribution aurora-agent
import aurora_agent
version 0.9.3
Python >=3.11,<3.14

Documentation

Independent verification challenges

AURORA publishes two OpenSSL-only RFC 3161 verification exercises. They require no AURORA SDK, API key, backend access, or custom verifier.

These challenges demonstrate cryptographic binding and tamper detection for the supplied bytes.

They do not establish capture completeness, legal authority, decision correctness, external-provider acknowledgement, or external-world truth.

What this SDK does not claim

The SDK does not claim:

  • capture completeness;
  • absence of bypass paths;
  • causal or external-world truth;
  • external-provider acknowledgement;
  • qualified timestamp status;
  • legal authority or legal compliance;
  • decision correctness, fairness, or policy validity.

DIGEST_ONLY protects raw values from the local outbox and AURORA payload storage, but AURORA can verify only the supplied client commitment unless the raw value is transmitted separately.

License

The aurora-agent SDK source and its generated Python distributions are licensed under the Apache License, Version 2.0. See LICENSE and NOTICE.

AURORA hosted services, the AURORA backend and frontend, and JAKROW components outside this SDK are proprietary and are not included in that grant. See LICENSE_HISTORY.md for the licensing history of releases before 0.3.0.

AURORA and JAKROW are marks of Krow Industries.

Metadata

Release files for aurora-agent 0.9.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for aurora-agent 0.9.3
File Size Uploaded
aurora_agent-0.9.3.tar.gz 104.7 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for aurora-agent 0.9.3
File Interpreter ABI Platform
aurora_agent-0.9.3-py3-none-any.whl Python 3 none any Details

Total release size: 172.1 kB

Release files / aurora_agent-0.9.3.tar.gz

Download URL aurora_agent-0.9.3.tar.gz
Size 104.7 kB
Tags Source
SHA-256 checksum
How to use checksums
9d4700c34c78e42489dc8569dec7d330447b283d448a5103224e88141d111166
BLAKE2b-256 checksum
How to use checksums
f3e3e7952b120c841a69173d74ea48f6e711d21be9d62cea1e1e9b9a4337c56a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 17, 2026.

Transparency log

Release files / aurora_agent-0.9.3-py3-none-any.whl

Download URL aurora_agent-0.9.3-py3-none-any.whl
Size 67.4 kB
Tags Python 3
SHA-256 checksum
How to use checksums
7f505f8d67675ba397cccce52a29af6b4c33aa34d7e8ed3b03e5f8c43ea5ec4f
BLAKE2b-256 checksum
How to use checksums
d54666481c643957476ba0ed632674bce5741bb949a88a81f807bf450a0c0a50
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 17, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

0.9.3 This release

2 release files

0.9.2

2 release files

0.9.1

2 release files

0.9.0

2 release files

0.8.0

2 release files

0.7.0

2 release files

0.6.0

2 release files

0.5.0

2 release files

0.4.0

2 release files

0.3.1

2 release files

0.3.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page