Tests As Behavioral Spec And Replication Surface

Issue 56 Edition 2026-02-25 6 min read

General

Sources: 1 • Confidence: High • Updated: 2026-04-13 03:42

Key takeaways

A comprehensive public test suite can enable a fresh reimplementation of an open-source library from scratch (including in a different programming language) by functioning as a behavioral specification.
tldraw filed a joke issue proposing translating its source code to Traditional Chinese as a purported defense against external AI coding agents replicating the project.
An issue that appeared to indicate tldraw would move its test suite to a private repository was later revealed to have been intended as a joke.
tldraw is described as not technically open source because its custom license requires a commercial license for use in production environments.
A tldraw maintainer suggested the project's defensible value lies in continually making strong product decisions for users rather than preventing others from recreating the code.

A comprehensive public test suite can enable a fresh reimplementation of an open-source library from scratch (including in a different programming language) by functioning as a behavioral specification.
The risk of test-driven reimplementation is presented as especially concerning for projects that combine open-source distribution with commercial business models.

tldraw filed a joke issue proposing translating its source code to Traditional Chinese as a purported defense against external AI coding agents replicating the project.
Cloudflare reportedly ported Next.js to use Vite in about a week using AI.

An issue that appeared to indicate tldraw would move its test suite to a private repository was later revealed to have been intended as a joke.
A tldraw maintainer argued that moving tests to another repository would complicate and slow development, and prioritized development speed over separating tests.

tldraw is described as not technically open source because its custom license requires a commercial license for use in production environments.

A tldraw maintainer suggested the project's defensible value lies in continually making strong product decisions for users rather than preventing others from recreating the code.

How often do independent implementations materially match a project's behavior primarily by relying on its public tests (without access to source), and how long does that typically take?
What were the precise scope, constraints, and verification criteria of the reported Cloudflare Next.js-to-Vite port, and what portion of the work was attributable to AI versus conventional engineering effort?
What exactly does the tldraw license require for production use (definitions, thresholds, enforcement), and how does that shape what competitors can legally do with reimplementations?
Are any serious projects (beyond jokes) actually moving test suites private or gating them, and if so, under what conditions (commercial pressure, AI-agent capability, compliance concerns)?
When teams consider separating or hiding tests, what measurable development costs (CI complexity, contributor friction, release cadence) show up in practice?

Public, comprehensive test suites can act as behavioral specs, making reimplementation easier and potentially weakening code-based defensibility for libraries whose monetization depends on proprietary-like control.
Operational costs of hiding or splitting tests may outweigh perceived IP protection, implying many teams may keep tests public even amid AI replication concerns.
Defensibility may shift toward licensing terms and continuous product decision-making rather than repository secrecy, especially for projects that are source-available with commercial production requirements.

Multiple independent reimplementations successfully match behavior using only public tests, with short time-to-parity and minimal access to original source.
Credible, non-joke examples of teams gating or privatizing tests specifically due to AI-assisted replication risk, with stated rationale and observable workflow changes.
Clear, enforced production-use licensing requirements materially constrain third-party commercial deployment, with reported licensing actions or widely acknowledged compliance practices.

Evidence that public tests rarely enable meaningful behavioral parity without source access, or that time and effort required remain prohibitive.
Teams that attempt to hide or split tests experience negligible development drag, suggesting operational cost is not a strong deterrent.
Licensing terms prove easy to route around via clean-room reimplementation and competitors successfully commercialize equivalent behavior without meaningful legal or practical friction.