Visual recaps.

Diagnostic deep-dive on the dns_bypass security signal: the 0.667 red card on the home network mission control dashboard, validation against Pi-hole's top-25 client list (17 of 26 covered), the two-root-cause split (public-fallback DHCP option 6 on Default vs full network isolation on VLANs 2 and 3), the deliberate resilience-vs-coverage tradeoff, and the ZBF policy chain at indices 10000 ALLOW, 10001 BLOCK, and 30000 Isolated Networks. Plus the Traffic Rules pivot story and the v0.4.0 release summary.

Twelve-slide narrative on resolving a dns_bypass finding from the home network mission control dashboard: validating the 0.667 metric, the resilience-vs-coverage design tradeoff on the Default subnet, the four-phase ZBF remediation (snapshot, ALLOW at 10000, DHCP rewire, BLOCK at 10001), the failed Traffic Rules attempt, the pivot to ZBF firewall-policies, and the public chris2ao/unifi-mcp v0.4.0 release with six new traffic_rules tools, an empty-body DELETE fix, and five schema discoveries.

A visual overview of the April 2026 Vercel breach response: the fourth-party supply chain from a Lumma Stealer infection on a Context.ai employee device through OAuth pivot into Vercel's internal tooling, the rotation ladder across nine exposed credentials, programmatic forensic verification against 100 deployments and 12 team events, and the hardening pass that locked the account down.

A slide deck walking through the 60-minute AI-assisted incident response: the anatomy of a fourth-party breach, the plan-file pattern for durable recovery checklists, the rotation ladder matrix sorting nine credentials by blast radius, programmatic truth (the UI lies, the API does not), the actionable hardening architecture across Vercel and Google Workspace, and why a SOC 2 badge is a claim rather than a proof.

A visual overview of building chris2ao/pihole-mcp: surveying five existing Pi-hole MCPs with /deep-research, consolidating three upstream repos into a single Python FastMCP server, 28 tools across six modules, the Pi-hole v6 session auth pattern with pre-expiry refresh and bounded 401 retry, and the open-source polish standard retrofitted to both public repos.

A slide deck narrating the chris2ao/pihole-mcp build: the deep-research phase, the fragmented upstream ecosystem, the consolidate-don't-extend decision, 28-tool modular architecture, the resilient Pi-hole v6 session auth pattern, the local DNS read-modify-write quirk, hermetic test design, the open-source polish standard, and the MCP vs. skill decision framework.

A visual overview of composing a custom UniFi MCP server from two open-source libraries: 103 tools across Network and Protect, the FastMCP closure trick, probe-by-status-code endpoint archaeology, PRODUCT_UNAVAILABLE stubs, and the single-repo plugin marketplace pattern.

A slide deck covering the full build of chris2ao/unifi-mcp: the /deep-research survey that found two contenders, the compose-don't-clone decision, three-loader lazy architecture, the FastMCP _bind_client closure, probe-by-status-code endpoint archaeology, AJV_PARSE_ERROR as a negative-space signal, PRODUCT_UNAVAILABLE stubs, and the Claude Code plugin install path.

A visual overview of the 5-agent UI/UX design team: the 3-tier research pipeline, director + specialist architecture, shared knowledge base, performance budget, security firewall, and the reskin vs. redesign lesson.

15-slide deck covering the end-to-end journey of building a 5-agent UI/UX design team: 12 skills evaluated in 3 tiers, the 36% prompt injection security decision, director + 4 specialists architecture, shared knowledge base, parallel quality gate, and engineering takeaways.

6-slide deck covering the three-tier research architecture: WebSearch limitations, Exa semantic search, Firecrawl JS scraping, the wrapper script security pattern, and the /deep-research skill workflow.

15-slide deck from a 22-source research report comparing persistent memory approaches for Claude Code. Covers the community landscape, the 5-layer architecture, Homunculus behavioral learning, fact versioning, and the memory-audit command.

A visual overview of the NotebookLM content pipeline: the 6-step generation process, security audit with three HIGH findings and mitigations, 4-dimension QA gate, and performance specs.

A 12-slide deep dive into building an autonomous content pipeline with NotebookLM: from threat modeling the unofficial API to the 6-step agent workflow, brand priming, and the 4-dimension QA loop.

A 10-page reference covering the full engineering approach to running an M4 Mac Mini as a 24/7 OpenClaw server: power management, network stall mechanics, watchdog cron implementation, and the midnight race condition.

A visual step-by-step guide to the three fixes that keep OpenClaw running 24/7 on a headless Mac Mini: power management, gateway watchdog, and daily log creation.

A deep-dive audio briefing on how popular MCP servers can leak Google credentials, the risks of community-built AI tool integrations, and how to evaluate them before installing.

A reference document covering the security risks of community MCP servers: credential extraction patterns, blast radius analysis, Google ToS violations, and a checklist for evaluating MCP tools before installation.

A reference guide covering the two-tier persistent memory architecture for Claude Code: global rule configuration, vector database MCP server setup, search modes, and troubleshooting.

A full visual map of the site architecture: Next.js 16 App Router, Neon Postgres, Vercel Edge, 20+ API routes, and all external services with icons.

A 10-slide visual recap of building cryptoflexllc.com from zero to production in one week with Claude Code.

A visual guide to installing Claude Code, what it does, and a 5-step roadmap for getting up and running.