TC39 Proposal Atlas

September 17, 2026 · View on GitHub

An interactive web application applying the TypeSafe AI System One paradigm (Jev) to the live corpus of ECMAScript / TC39 proposals.

Built with strict frontend design taste (tasteskill.dev), featuring a calibrated palette, deterministic typography, zero emojis, and an asymmetric split-screen architecture.

Design & Aesthetic (Taste Skill Guidelines)

  • Pastel Color Palette:
    • Stage 4 Finished: Pastel Mint (#DCF0E2 / #4E8664)
    • Stage 3 Candidate: Pastel Sky (#DBEAFE / #2563EB)
    • Stage 2.7 Spec Complete: Pastel Periwinkle (#E0E7FF / #4338CA)
    • Stage 2 Draft: Pastel Peach (#FDE4D2 / #A35C2B)
    • Stage 1 Proposal: Pastel Buttercream (#FDF3CD / #8A701A)
    • Stage 0 Strawman: Pastel Rose (#FCE0E8 / #9C3D5A)
    • Technical Domains: Soft Lavender, Sage, Peach, Mint, Rose
  • Anti-Emoji Policy: Strictly uses clean SVG primitives and Lucide icons.
  • Typography: Clean modern sans-serif (Plus Jakarta Sans / Geist) paired with high-clarity monospace (JetBrains Mono) for probabilities, ratings, and code tokens, with editorial serif display accents.
  • Layout & Viewport: Asymmetric split-screen layout with min-h-[100dvh] viewport stability.

Multi-Dimensional Semantic Analysis with TypeSafe AI

While standard records provide basic committee metadata (stage, name, description, authors), they lack structural and contextual nuance. TypeSafe AI enriches every proposal with a multi-dimensional semantic taxonomy evaluated using System One (Jev):

1. The 5 Core Dimensions

  • Adoption Path (Choice): Whether a proposal can be polyfilled in userland (userland_polyfillable), transpiled during build step (build_transpilable), or fundamentally mandates browser engine modifications (native_engine_required).
  • Cognitive Overhead (Score 1.0 – 5.0): The cognitive burden and mental model friction imposed on developers learning and reading code using this feature.
  • Primary Motivation (Choice): The core driving impetus: dev_ergonomics, performance_optimization, correctness_safety, or ecosystem_interop.
  • Web-Compat Risk (Score 1.0 – 5.0): The probability and severity of breaking existing websites or conflicting with legacy web APIs (e.g., Annex B, prototype pollution, de-facto library naming collisions).
  • Architectural Signals (Multi-Noul):
    • Mutates Prototype Chain: Does this modify native object prototypes or standard global constructors?
    • New Grammar / Syntax: Does this introduce new reserved keywords, operators, or syntax productions?
    • Cross-Realm Boundary: Does this cross or affect execution contexts, sandboxes, or microtask checkpoints?

2. Foundational Intent Archetypes & Historical Patterns

To uncover why proposals succeed or stall, TypeSafe AI classifies each proposal into one of 7 Foundational Intent Archetypes:

Intent ArchetypeHistorical ThesisShareStage 4 RateExamples
expressive_compositionSyntactic fluency, declarative flow, and eliminating boilerplate.44.1% (143)33.6%Pipeline Operator, Records & Tuples, Promise.any
standardizing_de_factoCodifying battle-tested ecosystem conventions from npm & browsers.15.7% (51)33.3%UUID, import.meta, Structured Clone, trimStart/trimEnd
correcting_historical_regretsRepairing 1995 Netscape compromises and legacy warts.15.1% (49)26.5%Temporal (replaces Date), Array Equality, Builtins.typeOf()
framework_enablementFirst-class runtime primitives and hooks for modern meta-frameworks.7.4% (24)16.7%Decorators, AsyncContext, Asset References, Module Keys
cross_language_parityAdopting proven constructs from Rust, Python, Ruby, C#, Java.6.8% (22)36.4%BigInt, RegExp Atomic Groups, Pattern Matching
hardening_integrityDefending against prototype pollution, realm escapes, supply-chain exploits.5.9% (19)0.0%ShadowRealm, SES, Freezing Prototypes, Defensible Classes
runtime_vitalityLow-level binary data, WASM interop, multithreading, zero-copy buffers.4.9% (16)37.5%SIMD, TypedArray Stride, ArrayBuffer transfer

Key Historical Insights:

  1. The Security Paradox (0.0% Stage 4): Proposals aimed at hardening_integrity face near-insurmountable friction reaching Stage 4. JavaScript's dynamic mutability makes locking down prototypes or isolating realms prone to web-compat breakage or severe committee deadlock.
  2. Proven Semantics Travel Well (36.4% - 37.5% Stage 4): cross_language_parity and runtime_vitality exhibit the highest consensus completion rates because the mathematical or systems semantics are already hardened in prior art.
  3. The Core Engine of JS is Expressiveness (44.1% Share): Almost half of all TC39 activity revolves around reducing developer boilerplate and enabling fluent chaining.

3. Interactive Evaluator Playground

Click any proposal to open the deep state inspector, view its exact immutable state JSON, explore full categorical probability breakdowns for its Intent Archetype, and run custom natural language conditions live in real time.

Running the App

cd ~/labs/tc39-typesafe # or cd ~/labs/typesafe-state

# Install dependencies
npm install

# Start development server
npm run dev

# Build production bundle
npm run build

# Synchronize dataset (fetches latest TC39 proposals & enriches new items)
npm run sync

Open http://localhost:5175 in your browser.

Automated Monthly Dataset Sync

A GitHub Actions workflow is configured in .github/workflows/monthly-sync.yml:

  • Schedule: Automatically runs on the 1st of every month at 00:00 UTC.
  • Manual Trigger: Can be dispatched on-demand via the GitHub Actions UI (workflow_dispatch).
  • Pipeline:
    1. Pulls the latest official TC39 committee archive.
    2. Runs npm run enrich:api to classify any newly submitted proposals or updated stages using TypeSafe AI (jev-latest).
    3. Verifies npm run build passes.
    4. Automatically commits and pushes updated proposals back to the repository if changes are detected.
  • Repository Secret: Set TYPESAFE_API_KEY in GitHub Repository Settings (Settings -> Secrets and variables -> Actions).