CCFish generates organic user acquisition by converting every development milestone into content: changelogs become blog posts, bug fixes become tutorials, and feature launches become case studies.
The Problem
Mobile game developers face a brutal math problem: paid user acquisition costs have risen 60% year-over-year, while organic discovery on the App Store grows harder with each algorithm update. For indie developers without VC funding, the standard playbook of buying users is simply not an option. CCFish needed a zero-cost acquisition channel that scaled with development output.
The Solution: Development-Driven Content
The insight was simple: every development activity generates collateral that can be repurposed into growth content.
| Development Activity | Content Output | Growth Channel |
|--------------------|----------------|----------------|
| Feature launch | Technical blog post | SEO + dev audience |
| Bug fix (interesting) | Tutorial / postmortem | Developer community |
| SDK upgrade | Migration guide | Partner ecosystem |
| Performance improvement | Benchmark comparison | Technical SEO |
| Architecture decision | Architecture deep-dive | Thought leadership |
Architecture: The Content Pipeline
CCFish runs a Cloudflare Workers pipeline that:
1. **Listens to GitHub webhooks** — new PR merged? Trigger content generation
2. **Classifies the change** — feature, bugfix, refactor, or infrastructure
3. **Generates draft content** — LLM-powered first draft based on the diff
4. **Queues for review** — stores in D1 for editorial review
5. **Publishes to blog** — auto-publishes via AIKit's D1 pipeline
```bash
The pipeline runs entirely on edge workers — no build server needed
1. GitHub webhook hits the worker
2. Worker queries D1 for template
3. Worker calls LLM via OpenRouter
4. Draft saved to D1
5. Cron job publishes next day
```
Step 1: GitHub Webhook Setup
```typescript
// cloudflare-worker-github-webhook.ts
export default {
async fetch(request: Request, env: Env): Promise<Response> {
const payload = await request.json();
if (payload.action === 'closed' && payload.pull_request.merged) {
// Queue content generation
await env.QUEUE.send({
type: 'pr_merged',
repo: payload.repository.full_name,
pr: payload.pull_request
});
}
return new Response('OK');
}
}
```
Step 2: Content Classification
Each PR is classified by LLM into one of the content types above. The classification determines the template and category for the generated post. For example, a PR titled "Add WebSocket reconnection logic" generates a tutorial, while "Reduce memory usage by 40%" generates a benchmark comparison post.
Step 3: Auto-Queue and Publish
Generated drafts land in the same D1-powered queue that the AIKit blog pipeline uses. A daily cron job picks them up, waits for editorial sign-off, and publishes. The entire pipeline from PR merge to blog post takes less than 24 hours.
Results
- **45+ posts generated** from development output in 3 months
- **70% of posts** required no editorial changes before publishing
- **2.4K monthly organic visitors** attributed to dev-to-content pipeline
- **Zero marginal cost** — uses existing LLM credits and Cloudflare free tier
Key Takeaways
The line between development and content creation is artificial. Every engineering team generates far more valuable content than they realize: changelogs, RFCs, pull request descriptions, architecture decisions. The key is building the pipeline that captures, formats, and distributes it.