---
id: agentrust-trace
version: "0.9.0"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# agentrust-trace — TRACE v0.2 — hardware-attested governance records for AI agents

License: permissive · Maintenance: active · Downloads: 78.7K/mo

## What it is and what it does

agentrust-trace is a Python implementation of TRACE v0.2, an open specification for hardware-attested AI agent governance records. It lets you create signed Trust Records that cryptographically prove what model ran, where it ran (in which TEE), under which policy, what data class it touched, and which tools it called—all bound into a single artifact rooted in silicon attestation. Any third party can verify the record without trusting the operator.

The package provides TrustRecord objects that capture agent execution context and sign them using cryptographic keys. It builds on IETF standards (RFC 9711 for CBOR Web Token/EAT, RFC 9334 for RATS roles, SCITT for transparency-ledger anchoring) and is designed for CoSAI WS4 interoperability. It's intended to integrate with agent governance frameworks like AGT and cMCP to create end-to-end compliance pipelines for confidential AI workloads.

Use it for:

- Prove to regulators or auditors that a financial AI agent ran under approved policy in a confidential environment without exposing operator infrastructure.
- Create verifiable compliance records for AI agents processing classified or sensitive data in regulated industries.
- Build a transparent governance pipeline where agent execution evidence is independently anchored to a ledger for non-repudiation.
- Integrate with confidential computing platforms (AMD SEV-SNP, Intel TDX) to bind agent behavior to hardware attestation.
- Enable third-party verification of AI agent behavior without requiring trust in the operator or access to internal logs.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Creates and verifies cryptographically signed Trust Records that prove an AI agent ran under a specific policy in a verified hardware environment, touching classified data and invoking identified tools.

Yes, if you are building AI agent governance infrastructure for regulated or high-assurance environments. The package is actively maintained, has no known vulnerabilities, and implements an emerging standard aligned with IETF and OASIS efforts. However, it is in Alpha (v0.2) and explicitly marked for developer preview—do not rely on it in production without reviewing the Limitations page. Install if you are prototyping or evaluating hardware-attested agent compliance; defer if you need stable, production-hardened tooling.

## Install

pip install agentrust-trace
uv add agentrust-trace
poetry add agentrust-trace

## Installing agentrust-trace

Before you install: Active development with a recent release (5 days old). Low install friction with pure-Python wheel distribution. Requires current Python versions (3.11+) and four standard dependencies (cryptography, jsonschema, pydantic, rfc8785). Early-stage project (Alpha status) launched June 2026.

License in practice: Apache-2.0 permissive license allows commercial use, modification, and distribution with minimal restrictions—suitable for proprietary agent governance pipelines.

Quickstart:

pip install agentrust-trace

from agentrust_trace import TrustRecord, sign_record

record = TrustRecord(
    subject="spiffe://trust.example.org/agent/payments-processor",
    model_id="claude-sonnet-4-6",
    platform="amd-sev-snp",
    policy_hash="sha256:b2c3d4...",
)
signed = sign_record(record, key=signing_key)

Requires Python 3.11 or later; signing and verification require cryptographic keys (not generated by the library).

Verify before relying:

- Whether the library handles SCITT ledger anchoring directly or requires external integration.
- Scope and maturity of the conformance test suite coverage.
- Production-readiness timeline and stability guarantees beyond v0.2.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 78.7K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

ai agent attestation, hardware-attested governance, trust record signing, confidential computing compliance, agent policy verification, cryptographic evidence records, tee attestation python, ai-governance, attestation, confidential-computing

[View on SkillFed](https://skillfed.io/packages/agentrust-trace) · [View on PyPI](https://pypi.org/project/agentrust-trace/)