Skip to content

CLAUDE.md

Latest commit

 

History

History
104 lines (86 loc) · 5.22 KB

CLAUDE.md

File metadata and controls

104 lines (86 loc) · 5.22 KB
Raw

Spaarke website — session start

This is the root routing doc. Before doing anything, identify what kind of work the session is about and route to the right place.

Domain dispatcher

If the work is… Read first
Writing or editing content (blog posts, LinkedIn posts, white papers, tweets, marketing copy) content-platform/CLAUDE.md
Site features, components, build config, infrastructure (no per-domain CLAUDE.md yet — read this file's §Site work below)
Bug fixes or maintenance (read this file's §Site work below)
Marketing operations (campaigns, calendar, GitHub Project) content-platform/CLAUDE.md + content-platform/github-setup.md

When the user's request is ambiguous, ask which domain before loading anything else. Loading the wrong domain wastes the context window and produces grounded-in-the-wrong-thing output.

Skills

The .claude/skills/ directory holds skills that enforce the gated workflows. Use them — they are not optional.

Skill When to use
idea-to-brief Writer has dropped an idea.md in content-platform/articles/<slug>/. Run this skill to produce brief.md.
content-pipeline A brief.md is signed off. Run this skill to produce plan.md + tasks.md + per-piece CLAUDE.md, create the GitHub Issue, add to the Project.
design-to-spec (software) Producing spec.md from a design idea.
project-pipeline (software) Producing plan + tasks from a spec.md.
task-execute (software) Working through a generated task.
push-to-github Push the current branch and open a PR.

If a skill exists for the work being asked, invoke it. Don't write a content brief from scratch when idea-to-brief is the codified, voice-aware, taxonomy-aware path.

Content-platform — quick orientation

Active operating program for everything Spaarke publishes. Lives at content-platform/. The structure mirrors the software project pattern — per-piece workspace with idea, brief, plan, tasks, CLAUDE.md, and draft.

content-platform/
├── CLAUDE.md                    ← writing-session router
├── spec.md                      ← architectural spec
├── voice/                       ← the constitution (style, brand, personas, vocab)
├── content-types/               ← per-type calibration (blog-post, linkedin-post, …)
├── templates/                   ← per-type workspace templates (copied when starting a piece)
│   ├── blog-post/{idea,brief,plan,tasks,CLAUDE}.md
│   ├── linkedin-post/...
│   ├── white-paper/...
│   └── tweet/...
├── articles/                    ← per-piece workspaces (actual pieces only)
│   └── <slug>/                  ← one directory per piece
│       ├── idea.md              ← writer's raw idea
│       ├── brief.md             ← formal brief (the spec)
│       ├── plan.md              ← outline + sections + sources
│       ├── tasks.md             ← workflow gates
│       ├── CLAUDE.md            ← per-piece session contract
│       └── draft.{mdx|md}       ← the working draft
├── campaigns/                   ← multi-asset coordinated pushes
├── calendar.md                  ← cadence record (with GitHub Project cross-ref)
├── github-setup.md              ← GitHub Project + Issues + Milestones layout
└── published/                   ← finalized linkedin posts, tweets, papers

The 16 already-published blog articles live at content/blog/ (the site's existing publish target). For new pieces, the workflow is: idea → idea-to-brief → content-pipeline → write → publish to content/blog/<date>-<slug>.mdx.

Site work — quick orientation

Until a per-domain CLAUDE.md exists for site code, here's the shortest loading list:

  • package.json + next.config.ts — Next.js 16.1.6 (Turbopack), Sharp, MDX
  • src/app/ — App Router routes
  • src/components/ — React components
  • src/lib/blog.ts — MDX frontmatter shape (load when touching blog frontmatter)
  • public/articles/<slug>/ — per-article static assets (heroes, LinkedIn headers)
  • Deploy: Azure Static Web Apps via GitHub Actions (.github/workflows/)

When site work grows enough to warrant its own routing doc (a docs/dev/CLAUDE.md or similar), add a row to the dispatcher above.

Repo conventions

  • Per-batch workflow: git fetch origin --prune → git checkout -b <type>/<slug> origin/main → edit → commit → push → gh pr create → gh pr merge --merge
  • Commit prefixes: content:, feat:, fix:, chore:, docs:. Content commits use the slug (e.g., content: <slug> -> outline).
  • Worktrees: the repo uses git worktrees for parallel work. c:\code_files\spaarke-website is the main worktree on main; c:\code_files\spaarke-website-wt-<feature> are branch-isolated worktrees. Create a worktree only when you need parallel work on multiple pieces; for one-piece-at-a-time, a regular branch is fine.
  • Don't push directly to main. Don't --force push. Don't skip hooks. Don't auto-publish content.