Write-upComing soon
Claude Code playbook
The exact workflow I use to ship production code fast: plan files, voice-locked CLAUDE.md, subagent use, verification checklist.
Claude CodeWorkflow
LABExperiments & notes
Where the tools get stress-tested before they hit client work.
Claude Code workflows, MCP server prototypes, agent experiments, and short technical write-ups. Everything here is shipped when it matters and abandoned when it does not.
Write-upLive
NEXUS, a self-hosted AI agent rig
Repurposed an older laptop as a standalone AI runtime. Linux Mint, Docker, OpenClaw, remote-managed over SSH and Tailscale, phone-accessible via a locked-down Telegram front-end. Kept off the main workstation on purpose.
LinuxDockerTailscaleSelf-hosted
Read write-up
Write-upComing soon
The AI agent decision playbook
When agents earn their keep, when a script does the job better, and the questions I ask before building one. Patterns from production work, not theory.
AI AgentsDecision-making
Write-upLive
Inside the Jumbo Gemini scanner
Production deep-dive on a shipped component: prompt design, confidence threshold, fallback to manual entry, and how the scanner sits inside the POS flow without blocking checkout.
GeminiVisionProduction
Read write-up
Two write-ups live, more in the pipeline. Start with the production AI vision deep-dive or the self-hosted agent rig.