💭 in your head · nothing to type
Picture an evening a month from now. You are on the road, messaging your bot from the phone: "What did I decide about moving, and why?"
A minute later the answer arrives — with the names of the notes it came from. All of it runs on your computer, in your files, with no new subscription.
Which piece of this picture do you already have, and which is still missing?
💭 in your head · nothing to type
You do not need a complete second brain at once. You need a minimal one that works: one memory, one channel, one skill.
Nobody can say in advance how long it will take. For some the build takes an evening, for others several evenings — it depends on experience and on what is already installed. Both are fine.
Think about what will be your sign that "it works": which question from the phone must get the right answer?
Now we connect the pieces from units 1–4 into a minimal working second brain.
1. Memory. The project from module 05 or the starter. It has notes/ with real notes, plus the "Second brain" and "Incoming messages" blocks in AGENTS.md.
2. Channel. First fakechat, then — if you want — Telegram following the Channels documentation. Only you are on the allowed list. The token sits in the channel's .env, not in the project.
3. Skill. One skill from someone else — the one you broke down in unit 3. Apply rule 1 from unit 4 to it. If everything is clear, copy it into your project's .claude/skills/<name>/. If something is unclear, pick another skill or build your own from a conversation.
4. Three checks.