Ideas from Modern Web Guidance

Is this an issue or a feature request?

Feature Request / Discussion

Description

The current Context Hub (chub) architecture relies on a reactive, community-driven workflow: developers encounter API shifts or hallucinations, generate feature requests, and the community builds corresponding markdown schemas.

Google’s new Modern Web Guidance (MWG) framework demonstrates a different, tightly integrated alternative. In MWG, the core platform engineers (such as the Google Chrome and Microsoft Edge teams) proactively author the guidance schemas.

We should explore expanding Context Hub to natively support "Vendor/Maintainer-Authored" tiers. This would allow platform creators—whether they manage browser APIs, language cores (Rust, Python), or major open-source libraries—to ship optimized agent instructions alongside their runtime code.

---

Architectural Comparison: Vendor-Proactive vs. Community-Reactive

Integrating a vendor-led model introduces an interesting set of trade-offs, particularly regarding time-to-market speed versus ecosystem breadth:

Pros of Vendor-Led Curation

• Zero-Day Agent Readiness (Faster Time-to-Market): Because a core platform engineer (e.g., a Python core dev or an internal library author) knows the feature inside out, they can submit the guidance schema before the code even launches. It completely eliminates the reactive lag of crowd-sourcing.
• Ultimate Ground Truth: The context comes straight from the spec authors, ensuring perfect technical accuracy and eliminating agent hallucinations about edge cases or complex runtime parameters.
• Proactive Upstream Updates: Guidance updates are tied directly to package release cycles and CI/CD pipelines, rather than waiting for a developer to break their build and report it.

Cons / Bottlenecks of Vendor-Led Curation

• Ecosystem Breadth (The Scale Bottleneck): While vendor-led is vastly faster for first-party features, it struggles to scale across the long tail of the ecosystem. Core teams cannot author guidance for millions of niche third-party libraries, community forks, or internal enterprise packages.
• High Maintenance Overhead: It requires continuous engineering effort from platform maintainers to ensure their hand-crafted instructions stay optimized for changing LLM context windows and behaviors.

---

Proposed Solution for Context Hub

Instead of picking one model, chub is uniquely positioned to merge both paradigms into a dual-tiered registry ecosystem:

1. chub get core/[namespace]: Proactive, zero-day schemas maintained directly by platform, language, and core framework teams (e.g., Python steering committees, Rust working groups, or major library maintainers).
2. chub get community/[namespace]: The existing reactive, crowd-sourced registry for the infinite tail of open-source packages and community-discovered workarounds.

By allowing authors to ship documentation explicitly optimized for AI context directly at release time, we can ensure coding agents hit the ground running with new language and library features from day zero.