9 content skills for Claude Code with persistent brand memory and an AI slop detector — plus a closed loop that studies your real results and rewrites its own guidance.
curl -fsSL https://raw.githubusercontent.com/quionie/claude-content-engine/main/install.sh | bash
or, inside Claude Code: /plugin marketplace add quionie/claude-content-engine
Every published piece gets logged. Run a retro — with connected analytics, or by pasting your numbers — and the engine compares winners against losers, keeps only findings with real evidence, and encodes them as drafting defaults. Your voice stays yours: the loop tunes tactics, never identity.
Ask naturally — Claude picks the right skill, loads your saved voice, and applies what past retros learned.
One piece becomes platform-native versions for Twitter, LinkedIn, email, and more. Re-angled, not resized.
SEO-structured posts with a differentiation angle — meta descriptions, heading structure, internal links.
Headlines, landing pages, and ad copy built on PAS, AIDA, and BAB frameworks.
Feed it your best writing; get a reusable voice profile any future session can follow.
Full drip campaigns — welcome, launch, onboarding, re-engagement — with timing and exit logic.
Weeks of scheduled, platform-formatted content mapped to your pillars.
Chains skills into pipelines: voice → blog post → repurposed set → posting calendar, in one run.
Your voice, audience, and preferences persist across sessions. Explain your brand once.
The learning loop. Grades your published content, extracts evidence-backed findings, updates the engine's own guidance.
Everyone learned to avoid "delve." The tell now is shape. The gate grades both, and Claude rewrites before you ever see the draft.
MIT · local-first, plain files · 30 tests on every push · every threshold is a named constant
Works with any Claude Code setup that supports plugins.
curl -fsSL https://raw.githubusercontent.com/quionie/claude-content-engine/main/install.sh | bash