Nothing is safe from a curious mind.

Posts in BLOG

  • I Was a Human AI Agent Before it was Cool  blog

    In 2026, what I'm about to describe would be a three-tool agent loop. An LLM instance with computer use, a couple of MCP servers, maybe a router for the edge cases. A weekend project. When I actual...  More »
  • Sending email from alias via Gmail  blog

    To send email from an alias address through your Gmail account, generate a Google App Password, then add the alias under Settings → Accounts and Import → Send mail as with smtp.gmail.com:587 as the SM...  More »
  • Git Branch Exists on Remote But Won't Show Locally  blog

    If a git branch shows up on the remote but git branch -r doesn't list it locally, your fetch refspec is almost always scoped to a single branch instead of all branches. Fix it with one config change: ...  More »
  • What Is MCP and Why Should Developers Care?  blog

    MCP (Model Context Protocol) is an open standard that lets AI models connect to external tools and data sources through a single, consistent interface. Anthropic introduced it in late 2024 to replace ...  More »
  • Fixing Godot MCP in Cursor on WSL  blog

    If godot-mcp won't connect in Cursor on WSL, the real culprit is almost always that Cursor is a Windows app trying to launch a Linux Node binary it can't see. The fix is to set wsl.exe as the command ...  More »
  • Multi-Model LLM Orchestration with OpenRouter  blog

    Multi-model LLM orchestration is the practice of routing AI requests to different models based on what each task needs — speed, cost, reasoning depth, or code quality. OpenRouter makes it practical by...  More »
  • Adding Semantic Search to Your Postgres App with pgvector  blog

    pgvector is a Postgres extension that adds vector storage and similarity search to an existing database, so you can run semantic queries directly against your application data without standing up a se...  More »
  • Building a Context-Aware AI Chat Without a Vector Database  blog

    You can ground an AI chat in your own data without a vector database by assembling the relevant documents directly into the system prompt before each request. No embedding pipeline, no similarity sear...  More »
  • How I Built an AI Document Ingestion Pipeline  blog

    Symport is an AI document ingestion pipeline that turns a phone photo of any paper document — receipt, EOB, prescription, utility bill — into structured JSON, then stores it in Postgres with embedding...  More »
  • Turning Manual Ops Into a 10-Minute Task  blog

    I once turned a 2-week manual data update process into a 10-minute automated pipeline by writing a PHP script that ingested a vendor spreadsheet, normalized everything into a temporary MySQL database,...  More »
  • Multi-Select in Visual Studio Code  blog

    TL;DR: I am suddenly using VS Code because of multi-cursor. Never thought I would. How the mighty have fallen. The Shortcuts (If That's All You're Here For) These are for Mac — they probably wor...  More »
© 2026 Sym