Open source · Free to fork

Your commits. In the feed.

devcast monitors every GitHub repository you own, analyzes each commit with 24 specialized modules, and generates LinkedIn + Instagram posts in your voice — automatically.

Use this templateView source

How it works

01

You push code

devcast watches every repository under your GitHub account. No configuration per repo — one setup, every push.

02

24 modules analyze it

Performance, security, architecture, Python, Go, Java, Elixir, React — modules run in parallel and surface what is actually interesting.

03

Claude writes in your voice

One Claude API call per commit. The post lands in Buffer as a draft. You review, edit if needed, and publish.

24 analysis modules

Modules run in parallel. Zero findings = no Claude call = no cost.

ComplexitySecurityPerformanceDesign PatternsClean CodeType SystemAPI DesignError ResilienceObservabilityConcurrencyPythonGoJava / QuarkusElixir / OTPReact PatternsDevOpsArchitectureTestingEvolutionaryDXAI-AssistedIntegrationDependency HealthJS Advanced

Built for the long run

Free tiers only

GitHub Actions, Buffer, and Supabase are free. Only the Claude API costs money — roughly $0.006 per post.

Human review always

Posts go to Buffer as Ideas. You approve, edit, and publish from Buffer's native UI. No auto-publish path exists.

Voice training loop

Every post you publish feeds back into the system. devcast learns your edit patterns and writes closer to your voice over time.

Extensible by design

Write custom analysis modules in TypeScript using the devcast-sdk. Publish to npm as devcast-module-*, list in config.

Up and running in 8 steps

  1. Fork the repo (Settings → Actions → Allow all actions)
  2. Create a Supabase project and run database/schema.sql
  3. Add 5 secrets: ANTHROPIC_API_KEY, BUFFER_ACCESS_TOKEN, SUPABASE_URL, SUPABASE_SERVICE_ROLE_KEY, CONFIG_YAML
  4. Copy config.example.yaml → set your GitHub username
  5. Run npm run setup-buffer to get Buffer profile IDs
  6. Edit voice-bootstrap.md with 3–5 posts in your voice
  7. Run npm run bootstrap to seed voice history
  8. Actions → "Poll and Generate" → Run workflow → check Buffer Ideas
Use this template