ProseForge
Full-stack content automation platform powering proseforge.ai
Content creators need a streamlined way to research, ingest, and generate stories from multiple sources. ProseForge is a writing platform, not a content automation framework.
- Built Go backend with Chi router, PostgreSQL, and River job queue
- Integrated AI story generation via Ollama/RunPod
- Created Vue.js frontend for content management
GoPostgreSQLVue.jsTypeScriptDockerTerraform
ProseForge Mobile
Native companion app for ProseForge on iOS and Android
Users need native mobile access to browse and read ProseForge stories.
- Built cross-platform app with .NET MAUI targeting iOS, Android, macOS
- Generated type-safe API client via NSwag/OpenAPI
- Implemented MVVM architecture with dependency injection
.NET MAUIC#XAMLOpenAPI
One More Story
One More Story by ProseForge, the companion game released with Codex and a Claude Agent Walk Into a Room...
The book Codex and a Claude Agent Walk Into a Room... needed a real app readers could install and play.
- Built a small idle-game engine in .NET MAUI where each world is data: rules in gameplay.json, words, art and voices in theme.json
- Shipped a medieval book guild plus other worlds on the same engine: Santa's Workshop, a village bakery, a rocket yard
- Tested the rules without phone tooling, and scripted uploads to Google Play and App Store Connect
.NET MAUIC#JSON themes
ProseForge Workbench
BYOAI client that lets any AI agent act as an author or reviewer on ProseForge
AI agents need a structured way to read, review, and improve stories on ProseForge without hand-rolling HTTP calls for every workflow.
- Built Go CLI and MCP server sharing a common service layer
- Generated typed API client from the ProseForge OpenAPI spec via oapi-codegen
- Released cross-platform binaries (darwin/linux × amd64/arm64) via GoReleaser
GoMCPOpenAPIoapi-codegenCobraGoReleaser
ProseForge TTS Worker
Pure-Go serverless TTS worker powering ProseForge audiobook narration
Narrating story audiobooks at scale needs a TTS pipeline that's fast, cheap, and doesn't drag a Python runtime into every deploy.
- Ported Kokoro-82M ONNX inference to pure Go — no Python dependency
- Built context-aware homograph disambiguation with Misaki dictionaries and Flite LTS fallback
- Deployed as RunPod serverless with zero cold-start and WAV/MP3 output
GoONNX RuntimeKokoro-82MRunPodFFmpegDocker