REPOSITORY ANALYSIS

tc39-atlas

A TC39 proposal explorer with adoption, complexity and compatibility rubrics, a local demonstration evaluator and a separate TypeSafe System One proxy.

E2 · source reviewedPublic material or source evidence was reviewed. A fixed-version source review requires checking a pinned revision; this is not a reproduction, performance endorsement, or safety guarantee.
Stars
1
Forks
0
License
—
Last commit
2026-09-17

What Jev does here

The server-backed route sends proposal state and Choice, Score and Noul questions to TypeSafe. The local typesafeEngine evaluator instead derives simulated values from keywords and rules.

Useful for comparing explicit proposal rubrics and inspecting state. Specification status remains a matter for TC39's primary records, not model scores.

The fixed local evaluator and HTTP proxy were reviewed. No live classifications were run; local demo distributions are not independently calibrated Jev outputs.

What task does this project fit?

Explore ECMAScript proposals through explicit adoption, complexity and compatibility rubrics.

Material analysed:

From input to action

  1. Build proposal state from its description and tags.
  2. Choose the local demonstration evaluator or the server-backed TypeSafe route.
  3. Inspect the selected dimensions and their provenance before using them in a decision.

Implementation details that matter

A simulated distribution is not a Jev result

typesafeEngine.ts labels itself a mimicking evaluator. It scores keyword overlap and normalizes handcrafted values; its numerical output is not an independently calibrated model probability.

Inspect the supporting source ↗

There is a separate live proxy

typesafeApi.ts has evaluate and classify-proposal handlers that POST to /v1/systemone. The existence of this path does not show that a particular displayed proposal was classified live.

Inspect the supporting source ↗

When to choose it

Useful as a taxonomy/playground for comparing rubrics. Use TC39 proposal text and stage records for standards facts; model scores do not determine specification status.

How to inspect it yourself

Compare evaluateChoice in the local engine with the server's classify-proposal request. Record which route produced the values instead of combining both into one benchmark.

Limits of this review

  • No server or model call was run by this site.
  • Risk and cognitive overhead are rubric-dependent judgments, not official TC39 measurements.
Primary material used in this analysis