Under the Hood
6 posts in this series
50 instincts, 13 semantic clusters, 7 accepted candidates, 5 promoted skills. I built the third tier of a continuous learning pipeline that synthesizes behavioral patterns into reusable agents, skills, and commands.
22 sources, 3 parallel research agents, 18 search queries. I pointed my deep research skill at the question every Claude Code power user asks: what's the best way to give an AI persistent memory? Here's what the community is doing, how my setup compares, and the 3 improvements I shipped the same day.
12 community skills evaluated, 35 design rules extracted, 4 knowledge base files created, 5 agents deployed. I built a complete UI/UX design and quality system for Claude Code in a single day.
247 game AI parameters, 7 candidate use cases, 5 agents, 1 honest verdict: no. But the research process itself uncovered three real configuration problems in my Vector Memory server that had been silently degrading search quality for weeks.
4 agents, 6 phases, 19 markdown files, 2 diagrams, 20 NotebookLM sources, 1 false positive caught, 1 silent UniFi bug surfaced and shipped as v0.3.0 in the same session.
5 existing Pi-hole MCPs. 1 actively maintained. 10 real gap items. I used /deep-research to scan the landscape, then consolidated 28 tools from three upstream repos into one Python FastMCP server that matches my UniFi MCP stack, and shipped it public with CI, issue templates, and branch protection.