[Feature Request] Enhance content creation and delivery via AI and automation

[nadvolod]

Is your feature request related to a problem? Please describe.

We have an opportunity to elevate:

• how we create content
• the quality of content
• our social promotion and communication strategy

Describe the solution you'd like

Some initial ideas are:

Priority 1: CLAUDE.md (~1 day) — High Impact

Create CLAUDE.md at repo root encoding:

• Temporal terminology rules from vale/styles/Temporal/terms.yml — Activity, Workflow, Worker, Namespace, Local
  Activity, Task Queue, Event History, etc. must be capitalized as proper nouns
• Common technical pitfalls learned from PR reviews:
  • "local Activity" = specific Temporal concept (Local Activity), don't use casually
  • HTTP calls must be wrapped in Activities, never called from Workflows
  • "shared deployment" is not a real limitation — teams can split into separate Workers on the same Namespace
  • Note feature availability (e.g., async updates not yet supported across Nexus)
• Style rules from STYLE.md: infinitive verb headings, sentence casing, "you/your" not "we/our", no headings deeper
  than H3
• Frontmatter template with required fields (id, title, description, keywords, tags, last_update, image)
• Tutorial structure (gold standard from Nexus tutorial): What you'll learn, Prerequisites, Scenario, progressive
  Checkpoints, Quiz, What You Built, What's Next + CTA
• Build commands: yarn start, yarn build, yarn format, yarn vale, yarn clearsnips

File: CLAUDE.md

---

Priority 2: Agents.md (~0.5 day) — High Impact

Create Agents.md (or .github/AGENTS.md) documenting the team's AI-assisted workflow:

• Which AI tools to use when: Claude Code for drafting/editing, CodeRabbit for PR review, Kapa.ai for doc search
• Review protocol: AI review first (CodeRabbit) → author self-review → lead engineer review
• Prompt patterns: How to ask Claude Code for tutorial drafts, technical reviews, comparison tables
• Quality loop: How AI feedback feeds back into CLAUDE.md and Vale rules (continuous improvement)
• Roles: What AI handles (terminology, structure, formatting) vs what humans handle (technical accuracy, narrative
  quality, feature correctness)

File: Agents.md

---

Priority 3: CI Pipeline (~1 day) — High Impact

Create .github/workflows/quality.yml triggered on PRs:

1. Vale prose linting — catches terminology violations, gerund headings, missing sections
2. dprint format check — catches formatting drift
3. Docusaurus build — catches broken links (config has onBrokenLinks: "throw")
4. Lychee link check on built output
5. Snipsync guard — grep to ensure no injected snippet content is committed

Run Vale + dprint in parallel; build then Lychee sequentially.

File: .github/workflows/quality.yml

---

Priority 4: CodeRabbit for AI PR Reviews (~0.5 day) — High Impact

Set up CodeRabbit on the repo:

• Configure .coderabbit.yaml with custom review instructions:
  • Temporal terminology rules (from CLAUDE.md)
  • Tutorial structure expectations
  • Feature accuracy checks (known limitations like async Nexus updates)
• Stronger than Copilot for prose/docs review — understands markdown structure, can reference style guides
• Runs automatically on every PR, posts inline comments
• Reduces lead engineer review load to technical accuracy and narrative quality

File: .coderabbit.yaml

---

Priority 5: Pre-commit Hooks + PR Template (~0.5 day) — High Impact

Activate Husky (already in devDeps):

• Add "prepare": "husky install" to package.json
• Create .husky/pre-commit running yarn clearsnips + npx lint-staged
• Configure lint-staged in package.json for markdown (dprint)

PR template (.github/pull_request_template.md):

• Checklist: Temporal terms capitalized, headings use infinitive verbs, Vale passes, build succeeds, feature
  accuracy reviewed

Files: package.json, .husky/pre-commit, .github/pull_request_template.md

---

Priority 6: Enhanced Vale Rules (~1 day) — Medium-High Impact

New rules in vale/styles/Temporal/:

• nexus-accuracy.yml — warn about async updates across Nexus, sync handler timeout limits
• terminology-context.yml — flag "local activity" (lowercase), "shared deployment", "in-process" near Temporal
  concepts
• feature-status.yml — flag references to known-limited features

These feed into both CI (Priority 3) and CodeRabbit (Priority 4).

Directory: vale/styles/Temporal/

---

Priority 7: Tutorial Scaffolding Script (~1 day) — Medium Impact

Create scripts/new-tutorial.js — interactive Node.js script (no extra deps):

• Prompts for: language, slug, title, author, editor
• Creates directory structure: docs/tutorials/{language}/{slug}/, images/, ui/
• Generates index.md from template matching the Nexus tutorial gold standard
• Add "new-tutorial": "node scripts/new-tutorial.js" to package.json

Files: scripts/new-tutorial.js, package.json

---

Priority 8: Copilot for AI PR Reviews (~0.5 day) — Medium Impact

Add GitHub Copilot PR review as a complement to CodeRabbit:

• Better for code-heavy tutorials (Java, Go, Python code blocks)
• CodeRabbit handles prose/structure, Copilot handles code correctness
• May be redundant if CodeRabbit covers enough — evaluate after Priority 4

---

Priority 9: CI Auto-Generate Comms (~2 hrs) — High Impact

GitHub Actions workflow triggered on merge to main:

• Detect new/updated tutorials (diff docs/tutorials/)
• Auto-generate draft communications:
  • Slack announcement for internal channels
  • Email digest snippet for newsletter
  • Social media posts (Twitter thread, LinkedIn post)
• Output as PR comment, Slack message via webhook, or artifact
• Needs careful prompt engineering to match Temporal's voice/tone

File: .github/workflows/content-promotion.yml

---

Priority 10: Social Posts Skill (~1 day) — Medium Impact

Manual-trigger version of Priority 9. Claude Code /social-posts slash command:

• Reads a tutorial file and generates Twitter thread, LinkedIn post, dev.to intro
• Stepping stone toward the automated CI version

File: .claude/skills/social-posts.md

---

Priority 11: Interactive Widget Framework (~2-3 days) — Medium Impact

Extract shared CSS/patterns from existing Nexus widgets into reusable templates. Create a /interactive-widget skill
for data-driven HTML widgets (matching games, drag-drop, before/after comparisons).

Files: static/html/shared/widget-base.css, .claude/skills/interactive-widget.md