I’m one person. No team, no funding, no DevOps hire.
I built an open-source CLI, published it to npm, and it watches 11 GitHub repositories while I sleep.
Breaking change in Claude Code? I get triaged before coffee. rust-analyzer ships a new binary format? A GitHub issue is auto-created with the affected files and migration steps already written.
No human in the loop until it matters.
This is what agentic development actually looks like for a solo developer — not the demo, not the polished conference talk, but the reality of building and operating a tool that has real dependencies you didn’t write.
The Stack
lspforge is a CLI that manages LSP (Language Server Protocol) configurations for AI coding tools. Language servers make tools smarter — they provide completions, diagnostics, and go-to-definition across editors. But every tool expects a different config format, and every language server has a different installation path. lspforge automates all of it.
The problem: lspforge touches 11 repos it doesn’t control. Claude Code, OpenCode, Copilot CLI, Kiro, Cursor, and six language servers. Any of them can ship a change that silently breaks my tool.
The Agentic Layer
Instead of manually monitoring those 11 repos every morning, I built an agentic ops layer:
- An n8n workflow checks all 11 repos daily at 9AM for releases from the past 3 days
- It sends each relevant release to Gemini 2.5 Flash with a focused question: does this break lspforge?
- Gemini cross-references the actual source code and either clears the check or creates a GitHub issue with severity, affected files, and migration steps
- I get a Telegram notification — either “All clear” or a prioritized action list
One person. Zero manual monitoring. The ceiling for solo developers has never been higher.
This Series
Over the next four posts I’ll break down the full stack:
- The vision — what it means to operate like a team as a solo developer (this post)
- The build — how lspforge was built using agentic development with Claude Code
- Maintaining it — what actually breaks after you ship, and why shipping was the easy part
- The automation — the exact n8n + Gemini monitoring stack, how it works, and what it costs
- The failures — what broke, how I debugged it, and what agentic development looks like when it isn’t working
Follow along if you’re building alone and want to operate like a team.