Agent reliability artifacts
Lint manifests, replay tool calls, normalize traces, and export portable run cards.
AuraOne Open / Reliability evidence
Create the evidence an AI release review actually needs.
Use source-inspectable CLIs, SDKs, Actions, and portable formats for agent traces, regressions, dataset cards, contamination checks, robotics quality, and technical review packets.
For engineering, evaluation, governance, and audit teams that need evidence to remain portable across repositories, CI systems, providers, and review workflows.
- License
- Package-specific; verify source, license, registry, and Action evidence before use.
- Data boundary
- Tools are intended to run in the operator's repository or CI boundary. Network, telemetry, artifact storage, and secrets depend on the selected tool and configuration.
- License
- Package-specific; verify source, license, registry, and Action evidence before use.
- Runtime boundary
- Tools are intended to run in the operator's repository or CI boundary. Network, telemetry, artifact storage, and secrets depend on the selected tool and configuration.
Product surface
The job, runtime boundary, and resulting artifact stay visible.
Inspect locally
Start from the source, then follow the reviewed README.
List the current public repositories first. Select a specific toolkit only after verifying its source, license, release, permissions, and support record.
gh repo list auraoneai --limit 100Evaluation and regression evidence
Version rubrics, protect regression banks, detect contamination, and publish decision-ready results.
Dataset and robotics records
Create dataset cards, rights references, quality checks, embodiment records, and reviewed release manifests.
AuraOne Trust Toolkit evaluation path
Source inspection, runtime testing, and release verification are separate checkpoints.
01
Select the artifact
Choose the exact reliability, evaluation, dataset, or release record the workflow requires.
Artifact definition
02
Inspect the implementation
Review source, license, dependencies, permissions, inputs, outputs, and configured network behavior.
Source and dependency review
03
Run inside the boundary
Execute against synthetic or approved data in the intended repository or CI environment.
Observed runtime record
04
Verify the artifact
Check schema, provenance, checksum, ownership, and compatibility with the decision workflow.
Validation report
05
Attach to the decision
Retain the exact artifact version and source beside the approval or release gate.
Evidence packet reference
Evidence packet preview
AuraOne Trust Toolkit evidence packet
The current release is trusted only when the source, artifact, runtime, and support records agree.
- Identity
- Tool, package or Action, version, owner, and source revisionSource and registry records
- Boundary
- Permissions, secrets, network, telemetry, and artifact storageConfiguration and observed run
- Output
- Schema, checksum, provenance, compatibility, and validationGenerated artifact
- Decision
- Workflow, reviewer, timestamp, state, and retained evidenceApproval or release gate
Release contract
A source link does not imply a verified package or binary.
| Field | Current evidence | State |
|---|---|---|
| Source | Organization and repository links are listed for inspection. | Available |
| License | Each repository, package, CLI, SDK, or Action requires its own verified license record. | Verify current release |
| Registry package or Action | Version, publisher, provenance, checksum, permissions, and compatibility require current evidence. | Verify current release |
| Support and security | Maintainer, disclosure path, update policy, and response boundary require project-specific verification. | Verify current release |