L · E · D · /led

The calibration framework with no specialist.

L-E-D is the only framework in Black Book that no specialist defaults to. That's deliberate. The other eight structure the work; L-E-D structures how the work feels.

What it is

• Level — Set the complexity level. Beginner / intermediate / expert? Junior PM reading this, or CTO?
• Expectation — What does the user want back, concretely? List of three? Two paragraphs? A single decisive recommendation?
• Direction — Style, format, tone. Plain and direct? Punchy and irreverent? Formal and grounded? Conversational?

When to reach for it

Most prompts don't need it. The framework on which the specialist is built (R-I-S-E, C-A-R-E, etc.) usually pins the work down enough.

Use L-E-D when the answer to "what does good look like?" depends entirely on stylistic choices the user has not made yet:

• A blog post on the same topic, for the same audience, could be a 600-word explainer or a 2,500-word deep-dive. → L-E-D forces the call.
• An exec comms message about a layoff could be sober and apologetic, or steady and direct. → L-E-D forces the call.
• A technical doc could be written for a junior on-call engineer or for a principal architect. → L-E-D forces the call.
• A sales email could be polite and professional, or irreverent and punchy. → L-E-D forces the call.

How to layer it

Pattern 1 — L-E-D in front of another framework

**Level:** Mid-career engineer, comfortable with distributed systems, hasn't run a database migration of this scale before.
**Expectation:** A runbook they could execute in 90 minutes with a buddy.
**Direction:** Calm, sequential, no breathless "this is critical" language. Tone is "here's how we do it; you've got this."

Now apply R-I-S-E:
**Role:** ...
**Input:** ...
**Steps:** ...
**Expectation:** ...

L-E-D goes first because the calibration shapes everything that follows. A hostile-takeover proposal and a friendly-merger proposal use the same C-A-R-E intake — but L-E-D distinguishes them.

Pattern 2 — /led standalone after a framework command

If you ran /rise or /care and the output is technically correct but stylistically wrong, run /led over it as a refinement pass:

/led make this read like a Stripe blog post — confident, plain, slight wit, no breathless launch language

This is the most common usage pattern. L-E-D is rarely the first call; it's usually the correction call after the substance is right but the temperature is off.

What it isn't for

• Don't use L-E-D when the user hasn't decided what they want — they need T-A-G or R-T-F to pin down the task itself first. L-E-D layered on a vague task makes the vagueness feel polished.
• Don't use it as a wrapper for "make it sound good." That's not calibration; it's begging. Specify what good means.
• Don't use it for length. "Make it 800 words" belongs in the framework intake, not in L-E-D.

Why it has its own slash command but no specialist

A specialist's framework should pin its intake. If we built a specialist around L-E-D, we'd be saying "this specialist's primary job is to calibrate tone." That's a niche role at best, and in practice it's a layer every specialist already does some of.

The slash command — /led — gets to be a focused tool: when you have a draft and the substance is right but the feel is wrong, run /led and get a calibrated rewrite. That's the use case, and it doesn't need a persona.