convez.ai
Skill
arome3

code-to-content

arome3/code-to-content

Ver no GitHub
BomPluginScore · de 0 a 100

Claude Code skill that transforms codebases into developer content. 5-phase gated workflow, 3 specialized agents, 13 slash commands. Generate blog posts, tutorials, Twitter threads, LinkedIn posts, READMEs, newsletters, video scripts, and conference talks—all grounded in actual code.

Instalação

$npx convez-ai arome3/code-to-content

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
7
Forks
4
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
v1.1.0
Issues abertas
2
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

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

FeatureDescription
5-Phase ProcessMandatory workflow: Analysis → Audience → Content → Optimization → Verification
Quick ModeReduced-phase commands for rapid content generation
Verification GatesEach phase has gates that MUST pass before proceeding
9 Content TypesBlog, tutorial, Twitter, LinkedIn, README, newsletter, video script, conference talk, product launch
Build-in-PublicJourney brief system for sustained narrative and voice continuity
Claude-NativeZero dependencies — all analysis performed inline
MCP ReadyOptional integration with social MCP servers for direct posting
18 Evaluation QuestionsSelf-testing suite to verify skill effectiveness

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]

Title: How we reduced latency by 40% using Rust (and why we ditched Python)

Intro: When we started the payment-service module (located in /src/services/payments), we hit a bottleneck. Our Python implementation was choking on concurrent requests.

In this post, I'll show you exactly how we refactored the AsyncProcessor class...

[...Full post continues with code snippets extracted from your files...]

Slash Commands

Full Mode (All 5 Phases)

CommandDescription
/c2c:blogGenerate a blog post
/c2c:tutorialGenerate a tutorial
/c2c:twitterGenerate a Twitter/X thread
/c2c:readmeGenerate a README
/c2c:linkedinGenerate a LinkedIn post
/c2c:newsletterGenerate a newsletter issue
/c2c:video-scriptGenerate a video script
/c2c:conference-talkGenerate a CFP + talk outline
/c2c:product-launchGenerate multi-platform launch content

Quick Mode (Reduced Phases)

CommandPhases SkippedBest For
/c2c:quick-twitter1, 2, 4Fast tweets when you know your audience
/c2c:quick-linkedin1, 2, 4Quick LinkedIn posts
/c2c:quick-blog1, 4Rapid blog drafts (still validates audience)
/c2c:quick-readme1, 2Fast README updates

What's Included

code-to-content/
├── .claude-plugin/
│   └── plugin.json             # Plugin metadata
├── skills/
│   └── code-to-content/
│       ├── SKILL.md            # Main skill definition (v1.0.1)
│       ├── .claude/commands/   # 13 slash commands (9 full + 4 quick)
│       ├── assets/templates/   # Content templates with copy-ready formats
│       ├── references/         # Claude-native analysis guides (13 files)
│       ├── examples/           # Positive, negative, workflow
│       ├── evaluation/         # 18 QA pairs
│       └── legacy/             # Archived Python scripts (optional)
├── README.md
└── LICENSE

The 5-Phase Process

PHASE 1 ──[Gate]──> PHASE 2 ──[Gate]──> PHASE 3 ──[Gate]──> PHASE 4 ──[Gate]──> PHASE 5 ──[Gate]──> DELIVERY
PhasePurposeGate Requirement
1. AnalysisUnderstand the codebase3+ content angles found
2. AudienceLock in ONE target audienceNo audience mixing
3. ContentGenerate evidence-based draftAll claims grounded
4. OptimizationApply SEO, voice, visualsVoice consistent
5. VerificationQuality gate checksAll thresholds pass

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 18 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
  • 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

Ver o original no GitHub

Bom para

Relacionados