# PactVerity Evidence Brief

Updated: 28 August 2026

Audience: independent reviewers, technical journalists, researchers and prospective integration partners.

## One-sentence description

PactVerity is a public AI-service evidence-verification project whose operational Proof Engine v1 and Receipt API beta convert caller-supplied service observations into deterministic, portable assurance receipts.

## What can be tested now

- Browser-based receipt creation, export and deterministic rechecking: https://pactverity.com/proof-engine
- Public Receipt API beta, health endpoint and OpenAPI specification: https://pactverity.com/developers
- Reproducible three-scenario benchmark and machine-readable fixtures: https://pactverity.com/benchmark
- Citation-ready JSON and CSV research dataset: https://pactverity.com/research
- Public integration acceptance gates and current register: https://pactverity.com/integrations
- Security disclosure policy and independent-review request: https://pactverity.com/security

## Latest first-party reference run

The 28 August 2026 live reference run produced the expected result in all three published scenarios.

- Result: pass, 3 of 3 expected scenarios
- Source commit: `7e8af9b4cb1b529a5c7462aa4f69208a6d1d6b13`
- Runtime: Node.js v24.19.0 on Windows x64
- Runner SHA-256: `0d17a8fc9e18bc5b4fd3c70a3d3c2091b76e50be9d301f841b7a2c8a8541ec27`
- Scenario fixture SHA-256: `00511eb9ac8283550489e120036a1ce24d23b2a049da183df06cc425d9d19e9d`
- Full machine-readable record: https://pactverity.com/research/benchmark-run-2026-08-28.json

This is a PactVerity first-party reference run. It is not an independent benchmark or audit.

## Current evidence boundaries

- Verified third-party integrations: 0
- Verified external benchmark reproductions: 0
- Independent audit: pending
- The benchmark tests deterministic evaluation of supplied observations. It does not prove that a caller's underlying observation is true.
- Planned custody, settlement and dispute layers are not yet deployed.
- Product operation does not prove customers, revenue, adoption, market liquidity or token value.

## Open independent-review requests

- Benchmark reproduction: https://github.com/PactVerityHQ/pactverity/issues/1
- API and SLA verification: https://github.com/PactVerityHQ/pactverity/issues/2
- Evidence adapter or integration proposal: https://github.com/PactVerityHQ/pactverity/issues/3
- Security and evidence-model review: https://github.com/PactVerityHQ/pactverity/issues/4
- Detailed review scope: https://pactverity.com/media/pactverity-independent-review-scope-v0.1.md

## Suggested independent tests

1. Run the three public fixtures from a clean environment and publish complete output, including failures.
2. Change one observation or criterion and verify that the receipt digest or check result changes deterministically.
3. Test malformed input, replay, substitution, oversize requests and misleading caller-supplied evidence.
4. Document the exact commit, runtime, commands, material relationship and limitations.

## Coverage guidance

Accurate coverage should distinguish working software, first-party test results, third-party reproductions, verified integrations, security reviews and future plans. An application, proposal, private demonstration or commercial discussion is not a verified integration. A first-party benchmark is not independent validation.

Technical and media enquiries: contact@pactverity.com