<p align="center"> <img src="c2c.png" alt="Code to Content" width="600"> </p>

<p align="center"> <a href="https://github.com/arome3/code-to-content/blob/main/LICENSE"> <img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="Code to Content is released under the MIT license." /> </a> <a href="https://github.com/arome3/code-to-content/issues"> <img src="https://img.shields.io/badge/contributions-welcome-brightgreen.svg?style=flat" alt="Contributions Welcome" /> </a> <a href="https://twitter.com/arome_dev"> <img src="https://img.shields.io/twitter/follow/arome_dev?label=Follow&style=social" alt="Follow on Twitter" /> </a> </p>

Code to Content

From commits to content.

Transform codebases into compelling developer content with a mandatory 6-phase process, verification gates, and build-in-public support.

Why Use This?

Most developers ship amazing code but struggle to tell the story.

• Writer's Block: Staring at a blank page is painful.
• Context Switching: Turning deep technical logic into accessible English is hard.
• Consistency: "Building in public" requires constant updates.

Code to Content solves this by using your actual source code as the source of truth. It doesn't just "write a generic post"—it analyzes your specific implementation, architectural decisions, and variable names to generate evidence-based content.

What This Skill Does

Installation

Option 1: Claude Code Plugin (Recommended)

/plugin marketplace add arome3/code-to-content
/plugin install code-to-content@code-to-content

Restart Claude Code after installation.

Option 2: Clone to skills folder

git clone https://github.com/arome3/code-to-content /tmp/code-to-content
cp -r /tmp/code-to-content/skills/code-to-content ~/.claude/skills/
rm -rf /tmp/code-to-content

Option 3: Copy to your project

# Clone and copy to your project's skills folder
git clone https://github.com/arome3/code-to-content /tmp/code-to-content
cp -r /tmp/code-to-content/skills/code-to-content your-project/.claude/skills/
rm -rf /tmp/code-to-content

Quick Start

Once installed, the skill auto-activates when you ask Claude to create:

• Blog posts or tutorials about code
• README, changelog, or API docs
• Twitter/X threads or LinkedIn posts
• Conference talk proposals
• Build-in-public updates

Example Usage

You: Write a blog post about my FastAPI project at ./src

Claude: [Executes 6-phase process]
  Phase 1: Analyzing codebase...
  Phase 2: Differentiation discovery (offer: the WHY, your opinion, roads not taken)...
  Phase 3: Locking audience...
  Phase 4: Generating content...
  Phase 5: Optimizing...
  Phase 6: Verifying quality gates + swap-the-name test...

  [Delivers verified content]

<details> <summary><b>See an example Blog Post output</b></summary>

Title: Why we rewrote our payment path in Rust — and the one service next door we deliberately left in Python

>

Intro (WHY-led, not a feature list): Everyone says "rewrite the hot path in Rust" like it's free. It isn't — you trade a language your whole team knows for a latency win you have to earn. The bottleneck was real: AsyncProcessor in payment-service (/src/services/payments) was serializing under concurrent load, and we measured it.

>

The road we didn't take: we kept the adjacent reporting-service in Python on purpose — it's batch, not hot, and the team velocity mattered more than the milliseconds.

>

[...continues with the trade-offs and code, grounded in your actual files — then a Phase 6 swap-the-name check confirms a competitor couldn't republish it...]

</details>

Slash Commands

Full Mode (All 6 Phases)

Quick Mode (Reduced Phases)

What's Included

code-to-content/
├── .claude-plugin/
│   └── plugin.json             # Plugin metadata
├── skills/
│   └── code-to-content/
│       ├── SKILL.md            # Main skill definition (v1.1.0)
│       ├── .claude/commands/   # 14 slash commands (9 full + 4 quick + cascade)
│       ├── assets/templates/   # Content templates with copy-ready formats
│       ├── references/         # Claude-native analysis guides (14 files, incl. differentiation.md)
│       ├── examples/           # Positive, negative, differentiation, workflow
│       ├── evaluation/         # 25 QA pairs
│       └── legacy/             # Archived Python scripts (optional)
├── README.md
└── LICENSE

The Differentiation Engine

AI made competent prose free, so grounded but generic now gets ignored — cover the logo, swap in a competitor's name, and most "We're excited to announce…" posts read identically. v1.1.0 adds the layer that fixes this:

• Lead with the WHY, not a feature list — the thesis your work is a consequence of.
• Hold one defensible opinion you'd stand behind, instead of committee-hedged mush.
• *Show what you chose not to build* — the trade-offs and rejected paths only you know.

Phase 2 gathers this from you — a short interview, raw material (Slack threads, notes, a voice-memo transcript), or a rough draft to polish ("mess over spec"). Phase 6 then scores the result 0–5 and runs the swap-the-name test. It never blocks you; it just refuses to pretend code-only output is distinctive, flagging it AT RISK instead. See references/differentiation.md.

The 6-Phase Process

P1 ─[Gate]→ P2 ─[Gate*]→ P3 ─[Gate]→ P4 ─[Gate]→ P5 ─[Gate]→ P6 ─[Gate]→ DELIVERY

_{Phase 2's gate is soft* — it never blocks delivery. Decline the interview and the skill proceeds on code alone, flagging the result distinctiveness AT RISK.}

Build-in-Public Support

For sustained content creation, the skill includes a Journey Brief system:

• Persistent context across content sessions
• Milestone-to-content mapping (what to post when)
• Voice continuity checklists
• Content cascade timing (Twitter → LinkedIn → Blog → Newsletter)

See references/build-in-public.md for the full guide.

Evaluation

Test skill effectiveness with 25 self-contained questions:

# In Claude Code with skill loaded
claude

# Ask any evaluation question, e.g.:
> According to SKILL.md, what is the maximum jargon density for intermediate developers?
# Expected: 4%

Categories tested:

• Template structure knowledge
• Audience profile thresholds
• Voice calibration rules
• Script capabilities
• Reference file content
• Content quality gates
• Differentiation (Phase 2 + swap-the-name test)
• Build-in-public workflow

Requirements

• Claude Code — No external dependencies required
• All analysis is performed inline using Claude-native protocols
• Optional: MCP servers for direct social media posting (see references/mcp-integration.md)

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Add/update evaluation questions if adding features
4. Submit a pull request

License

MIT License — see LICENSE file.

Author

Abraham Onoja · @arome_dev