Add v0.2.0 documentation and update website for release

[smintz]

This PR adds comprehensive documentation for protoconf v0.2.0 and updates the website to reflect the new release.

Summary

Adds new documentation pages covering v0.2.0 features, updates existing docs with v0.2.0 changes, and configures the Docusaurus site to support versioned documentation with v0.2.0 as the current version and v0.1.7 as a legacy version.

Key Changes

New Documentation Pages:

• docs/cli-reference.mdx - Complete CLI configuration reference for all commands (agent, serve, compile, insert, mutate, mod) with flags, environment variables, defaults, and descriptions
• docs/whats-new-0.2.0.mdx - Feature overview and upgrade guide from v0.1.7, highlighting breaking changes in configuration precedence and telemetry defaults
• docs/production/kubernetes.mdx - Guide for running protoconf on Kubernetes using the new ConfigMaps backend
• docs/production/rollouts.mdx - Documentation for staged rollouts feature with examples and observability
• docs/production/observability.mdx - Prometheus metrics and OpenTelemetry configuration guide
• docs/advanced-usage/modules.mdx - Remote module dependency management with CONFIGSPACE and protoconf.lock
• docs/advanced-usage/formatting.mdx - protoconf fmt usage and CI integration
• blog/2026-09-03-protoconf-0-2-0.mdx - Release announcement blog post

Updated Documentation:

• docs/consume-config-updates.mdx - Added GetConfig RPC and HTTP endpoint documentation, updated service definition with v0.2.0 changes
• docs/validation.mdx - Added protovalidate constraints in proto files alongside existing Starlark validators
• docs/getting-started.mdx - Added container image installation, updated sidebar position
• docs/production/running-in-production.mdx - Added ConfigMaps backend reference
• docs/production/inserting-configurations.mdx - Updated to mention ConfigMaps backend

Site Configuration:

• Updated docusaurus.config.js to set v0.2.0 as current version and v0.1.7 as legacy version
• Added versions.json to track versioned documentation
• Updated homepage link to point to v0.2.0 docs
• Adjusted sidebar positions across documentation hierarchy
• Updated homepage buttons with version-specific links

Minor Updates:

• Fixed typo in versions.json filename (was vesrions.json)
• Updated MDX React dependency to v3.0.0
• Added flex-wrap and gap to homepage button styling
• Updated README.md Docusaurus reference

https://claude.ai/code/session_01JvFenuqwiFW5vS3kf1Ezo8

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	www-protoconf-dev	55381a6	Commit Preview URL Branch Preview URL	Sep 14 2026, 07:40 AM

[smintz]

Workers Builds: www-protoconf-dev is failing, and it isn't this PR's

Standing down on this check rather than pushing a fix, because the failure is a project-configuration issue rather than something in this branch. The evidence:

• It fails instantly. started_at and completed_at on the check run are both 03:15:45 — zero seconds. It never reached an install or build step, so nothing in the diff was exercised.
• It didn't run on the first push. The check is absent on 32e1689 (03:53 UTC) and appears on 24088bc (03:15 UTC), so the Cloudflare Git integration was connected between the two.
• The only change between those commits is two lines in blog/2023-05-16-new-site.mdx — a {/* truncate */} marker in a Markdown post. There's no mechanism by which that breaks a Workers build.
• The repo has no Workers configuration. No wrangler.toml or wrangler.jsonc, no Workers entrypoint, and no static-assets binding anywhere in the tree. This repo currently deploys through .github/workflows/deploy.yml to GitHub Pages, with CNAME carrying the custom domain.

Test deployment — the GitHub Actions check that actually builds this site with yarn install --frozen-lockfile && yarn build — is the one that validates this branch, and it passed on the previous head.

What would fix it, if the intent is to move hosting to Cloudflare: the Workers project needs a build command of yarn build, an output directory of build, and a wrangler.jsonc declaring the static assets, roughly:

{
  "name": "www-protoconf-dev",
  "compatibility_date": "2026-09-03",
  "assets": { "directory": "./build", "not_found_handling": "404-page" }
}

I've deliberately not pushed that. Choosing between GitHub Pages and Cloudflare Workers — and what happens to CNAME, the existing Pages deploy workflow, and the custom domain — is a hosting decision for a maintainer, and I have no Cloudflare access to verify such a config would actually deploy. Adding it blind would widen a docs PR into a hosting migration.

I also can't re-run this check: it's posted by the Cloudflare app rather than GitHub Actions, so the usual re-run isn't available to me. Happy to add the config in a separate PR if you want to go the Workers route, or this check can be made non-blocking until the project is configured.

---

Generated by Claude Code

[smintz]

Correction to the timestamp in my comment above: the first push (32e1689) was at 02:53 UTC, not 03:53 — I typed it wrong, which made the ordering read backwards. The sequence is 32e1689 at 02:53 with no Workers check, then 24088bc at 03:15 where it appears and fails in 0s. The conclusion is unchanged.

---

Generated by Claude Code

[smintz]

Update: config pushed, and it did not fix the Workers build

Superseding part of my earlier comment. At the maintainer's request I pushed wrangler.jsonc in 784f193 (static assets from ./build, drop-trailing-slash to match trailingSlash: false, not_found_handling: "404-page"), verified locally with wrangler deploy --dry-run reading all 267 built files.

The build failed again on 784f193, again in 0 seconds (started_at and completed_at both 03:25:01) — identical to the failure before the config existed.

That result corrects my earlier diagnosis. I attributed the failure to the missing wrangler.jsonc; adding it changed nothing, so the missing config was not the proximate cause. A 0-second failure means the build ends before repo content is read at all, which puts the problem in the Cloudflare project's own configuration rather than in this branch.

I can't narrow it further from here. The check run exposes no log text to GitHub — output.text is empty and everything lives behind the dashboard, which I have no access to. The remaining candidates all sit on the Cloudflare side: no build command configured on the project, an incomplete Git-integration setup, or an account-level block on Workers Builds.

What would settle it: open the build log`` and read the error. If it names something fixable in the repo, I'll fix it.

wrangler.jsonc is still correct and needed for the migration regardless, so it stays. The GitHub Pages workflow and CNAME remain in place, so the live site is unaffected either way.

---

Generated by Claude Code