Part 2 of 3. Render the second brain vault as a printed daily brief each morning at 05:00 and deliver it to a reMarkable Paper Pro through a self-hosted rmfakecloud, with no reMarkable cloud involved.
read moreOther articles
Giving AI agents memory
Part 1 of 3. A markdown vault and a small MCP server that give any AI agent persistent memory, an action list with derived priority, and a calendar of one-off and annual events.
read moreSkill or MCP tool? A rule of thumb for Claude integrations
A practical rule for deciding whether a new Claude capability should be a skill (prompted procedure) or an MCP tool (deterministic lookup/action), worked out while adding a "daily knot" feature to a SecondBrain morning briefing.
read moreA Second MCP Server: Managing a Pelican Blog from Claude
How to add a second self-hosted MCP server alongside an existing one — a blog-management MCP that lets Claude create posts, add images, and publish a Pelican site without leaving the conversation.
read moreBuilding a Personal MCP Server for Your Second Brain
How to build a self-hosted MCP server that exposes a markdown vault as Claude tools — list projects, search notes, write entries, and commit changes, all from a conversation.
read moreAdding OAuth Authentication to a Self-Hosted MCP Server
How to add OAuth 2.1 authentication to a self-hosted MCP server: nginx as an authenticating reverse proxy, a minimal Node.js OAuth server, and the key gotchas that will trip you up.
read moreFSM Diagrams in Pelican and Claude Code
Finite state machines turn up everywhere: protocol implementations, hardware controllers, UI flows, parsers. Drawing them well is useful but tedious …
read moreVerilog Lint Skill for Claude Code
I've been using Claude Code for RTL generation lately, and the missing piece was a tight feedback loop between code …
read moreWaveDrom Timing Diagrams in Pelican with Claude Code