Frameworks

The Foundation hosts three open standards and operates one practitioner credential. Each open standard is independent: each can be adopted on its own, without committing to the others. The credential certifies competence in applying the methodology and is open to any practitioner.

Taken together, they answer one question: how to make software safe at the point where its behaviour is determined, not only at the end where failures are caught. AI now writes a large and growing share of the world's code, which makes the quality of that code a safety question. The Honest Framework constrains how code is built so that whole categories of defect cannot arise in the first place, including in code an AI generates. The Slop Audit then measures, independently and reproducibly, how much of any codebase lies beyond the reach of testing at all: the evidence that shows where construction is needed and whether it worked. MÉTRON works on a different source, measuring how the language a model is trained on shapes what the model can do, so capability can be anticipated from its inputs rather than discovered only after release. And because tools change nothing without people who can use them well, the Certified Honest Practitioner credential builds that competence. You cannot inspect quality or safety into a system after it is built; you build it in from the start.

Honest Framework

Honest is a specification for writing code that is correct by design. It is language-agnostic: any object-oriented or imperative language can implement it. The patterns and rules are the same; only the syntax changes.

In most cases you can also keep the application framework you already use within your language: Rails in Ruby, Django or FastAPI in Python, Spring in Java, Express in Node, ASP.NET in C#. Honest layers on top of these, not in place of them. You do not have to rewrite your application to start using it.

What does Honest actually require? In the simplest terms, two rules. First: write your decision logic as tables (rows of conditions, columns of outcomes) instead of as nested if-then-else chains and loops. Second: any single location in memory can only be changed by exactly one line of code in the whole system. These two rules together stop your program's possible states from multiplying out of control. They give you code you can check by reading instead of code you can only test by running every possible case.

Most ways to write provably correct software require you to learn a new language (Haskell, Idris, Coq) or to fight with complex type systems. Honest does not. You stay in your normal language and your normal framework. The specification gives you patterns and conformance rules that make your code checkable by reading its structure, instead of by trying to test every possible state.

Honest is general-purpose. It applies anywhere you write code that has to be correct:

Status:

Slop Audit

The Slop Audit is the Foundation's measurement instrument for software quality. It scores any codebase against the published thresholds of named compliance frameworks (SOC 2 Trust Services Criteria, NIST SP 800-53, OSFI B-13, OWASP ASVS, ISO/IEC 25010, WCAG 2.2, Section 508, EN 301 549, AODA, and Quebec Law 25) across 18 dimensions covering security architecture, data architecture, compliance engineering, operational security, performance engineering, operations, DevOps, infrastructure, software architecture, governance, process engineering, lifecycle management, and software development.

You can't eyeball quality. Slop is what the audit finds, not what the surface suggests. The word gets used as an impression: code that looks AI-generated, rushed, or low-effort. The Slop Audit replaces the impression with a measurement, an evidence-based, reproducible score against named indicators and compliance thresholds. A polished codebase can score badly, and rough-looking code can score well. Only the structure decides.

The Slop Audit is independent of the Honest Framework. It applies to any codebase, in any language, regardless of architectural style. The audit produces an evidence-based, reproducible score, with each indicator calibrated against the compliance-framework thresholds it maps to; the Honest Framework is one rigorous way to pass the audit by construction, but other architectures and methodologies can pass too.

What the audit measures:

The first cross-language measurement of one Layer 1 indicator (the mutable-state ratio) on 200 public open-source codebases finds that approximately 99% of non-React enterprise codebases are structurally incapable of exhaustive behavioural verification. The methodology and underlying paper are published under the Open Honest research program.

Pre-registered research and publications:

Status:

MÉTRON Framework

MÉTRON is a measurement-instrument family for cross-linguistic AI research. It comprises trained model checkpoints (currently MÉTRON-FR; a 12-language family in preparation) and, in active development, a no-code platform that lets researchers run controlled cross-language experiments without ML expertise or expensive hardware. You set up the experiment in a web interface; the platform runs it on dedicated GPUs at low cost.

MÉTRON also includes a community channel where native speakers of any language can build their own grammar test sets and get free compute in exchange for sharing those test sets back to the open release.

Pre-registered research and publications:

MÉTRON makes this method usable by people who are not ML engineers: philosophers of language, theologians, computational linguists, cognitive scientists, and cross-tradition scholars.

Status: methodology proven by pre-registered, peer-reviewed research; the no-code platform is in alpha development. Target platform release: EMNLP 2026 system demonstrations track (submission August–September 2026). Released under Apache-2.0.

Certified Honest Practitioner

The Certified Honest Practitioner credential attests to demonstrated competence in applying the Foundation's methodology: rigorous software-quality measurement using the Slop Audit, the construction-time discipline of the Honest Framework, and (where applicable) the cross-linguistic research methodology of MÉTRON. The credential is awarded on the basis of measured competence, not membership; it is open to any practitioner who demonstrates the required competence regardless of institutional affiliation, geography, or prior credentials.

The certification curriculum is the Foundation's own curriculum, written for the Foundation's certification program. It is independent of any commercial training material and can be completed without purchasing any commercial product. The credentialing program operates at cost-recovery candidate fees only, comparable to the candidate-fee structure of recognized 501(c)(3) credentialing organizations.

Status: curriculum and assessor-training procedures in development. The Foundation's practitioner-certification program is structurally comparable to the credentialing programs operated by IEEE, the Project Management Institute, ISACA, ISC², and the credentialing programs of the American Bar Association sections.