Why does naming chats matter?

You dispatch a new Claude chat to do a piece of work almost every day. Each one gets a name. Right now those names sort of follow a pattern. They don't have to. This lesson explains what we're trying to fix.

Lesson 1 — The problem
Concept 01 — Dispatching

You start a new chat almost every day.

You don't do all your work in one giant Claude conversation. You start a new chat to research one thing, a different chat to build another thing, a third chat to coordinate them, and so on. Each chat focuses on one piece of work, finishes it, and shuts down.

That's called dispatching — like a 911 operator sending out a police car. Each car gets a job, drives there, finishes it, comes back. You send each chat out the same way.

🧑 You AL-1.3 naming research SL-0.6 orchestrator P-3 scaffold-project 02.3.2 migration drain
one of you · many chats · each with a short tag

Each chat needs a name short enough to refer to in conversation. "SL-0.6 has an update" is faster than "the sixth orchestrator baton-pass in the skills lab." That short name is what we're trying to lock down.

Concept 02 — Today's shape

Right now a name has three parts.

You've already arrived at a pattern without writing it down. Let's break it open and look at the pieces.

SL - 0 . 6 WORKSHOP "skills lab" CATEGORY "0 = orchestrator" WHICH CHAT "the 6th one" "skills-lab, orchestrator category, 6th chat"
three parts · one job each

The workshop says which folder this chat belongs to. AL is the parent (Agentic Lab). SL is skills. RL is rules. Each workshop is its own world.

The category says what kind of work this chat does. In your AL workshop, 0 means orchestration, 1 means research, 2 means build, 3 means audit. You decided this once at the start.

The which-chat number is just a counter — the 6th chat of that category, the 7th, the 8th. Pure sequence.

The vocab

The short name itself (like SL-0.6) is called a tag. Like a license plate for the chat. When I say "tag" anywhere else, that's what I mean.

Concept 03 — What's drifting

Three small things keep quietly going wrong.

The pattern works. Nothing's broken. But three small things keep happening, and each one is friction. Read them so the rest of the course makes sense.

Drift #1 — em-dashes in titles are hard to dictate

Most of your chat titles look like this:

today (~75% of titles)
# SL-0.6 — Title

That long dash is an em-dash. Looks great. Hard to type. Your phone doesn't dictate it. Whisper just bails and writes a regular hyphen. Costs Claude a little extra every time.

proposed
# SL-0.6: Title

Colon. Easy to dictate ("colon"). Easy to type. Looks fine. Reads the same.

Drift #2 — two ways to spell one thing

In the settings header at the top of your handoff files, half of them say:

half the files
chat_name: SL-0.6

Underscore. Matches your other field names (parent_session, session_id).

other half
chat-name: SL-0.6

Hyphen. Doesn't match the others. If you ever wrote a script to read "all my chat names," it'd have to handle both.

Same information. Two spellings. The fix is just picking one and using it forever. (You don't have to know which one yet — we'll bless that in the walk.)

Drift #3 — nobody knows what "SL-0" means

If you tell a brand-new Claude chat "SL-0 has an update," that chat doesn't know what SL-0 is. The information exists — it's scattered across three different rule files — but no single place lists it. Like having a building directory split between three drawers.

The fix

One small directory in _agentic-lab/CLAUDE.md that lists every workshop prefix + what its numbers mean. Plus a one-line-per-chat log so a fresh session can grep "what was SL-7.1.1 again?" in two seconds.

Concept 04 — The goal

Lock what works. Fix the three small things. Add a directory.

That's the entire mission of Naming 101. We're not redesigning the system — your existing pattern is 90% right. We're writing down the parts that are unwritten, fixing the three drifts, and making sure new chats start the right way.

The hard work is already done. You've been doing it for months. This course just makes the unwritten rules visible so future-you (and Josh, when he joins) don't accidentally reinvent them.

What's in the next 3 lessons

Lesson 2 — Where the names actually live. (Handoff file frontmatter vs Claude Code's own session file. You asked me this; here's the long answer.)

Lesson 3 — Two new ideas we're considering. A "mission" layer and how to handle multiple seat types in one major.

Lesson 4 — Where we are now. What's locked, what's still open, what comes next.

Recap of Lesson 1

What you now know