Skip to content

yuvalraz/clutch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

clutch

The coupling between intention and action.

Clutch is an execution harness for ADHD. Barkley's frame: the problem isn't knowing what to do or wanting to — it's the wiring between intention and action at the point of performance. So Clutch isn't a motivation supply. It's a transmission: it gets your existing intent to the wheels.

Its foundation, and the first thing built here, is a weighted, queryable library of the ADHD research that actually resonates — one source at a time, in public. The point isn't the finished library; it's that it can't be abandoned: each source is small, self-contained, and shipped the day it's added. If the tool meant to help me finish things can't finish itself, it doesn't work. This repo is that first test.

The one rule (walking-pace floor)

One source → one clean file → one commit. Never "ingest everything, ship when done." The streak can't hit zero: the smallest legal move is adding a single source. Add the next one when the streak needs feeding — not before.

How it's stored

Plain markdown, one file per source in sources/, with frontmatter:

author: Russell Barkley
title: 30 Essential Ideas 1A — Intro
type: lecture          # lecture | book | article | podcast
source: https://...
resonance: high        # high | med | low — how much it maps to MY condition

resonance is the "weight" — hand-assigned, not learned. No vector DB yet: a folder of markdown is the database (grep + read reason over it directly).

Raw transcripts stay local (they're copyrighted third-party text). What's public is the tool, the index, and the resonance weights — not the source text.

Ingest a source

python3 -m venv .venv
.venv/bin/pip install -r requirements.txt

# a single video:
.venv/bin/python ingest.py "https://www.youtube.com/watch?v=VIDEO_ID" \
    --author "Author" --title "Talk title" --resonance high

# a whole playlist → ONE digest (a lecture series is a single work):
.venv/bin/python ingest.py "https://www.youtube.com/playlist?list=PLAYLIST_ID" \
    --author "Russell Barkley" --title "30 Essential Ideas (full series)" --resonance high

The canonical spine (ship one coherent source at a time — NOT all at once)

A source is a whole work. A lecture series chopped into parts is one source — digest it as one file, not N stubs.

  • Barkley — 30 Essential Ideas You Should Know about ADHD (full 27-part lecture → one digest)
  • Barkley — ADHD and the Nature of Self-Control (1997)
  • Barkley — Executive Functions (2012)
  • Thomas E. Brown — Smart but Stuck (the high-IQ coast-to-collapse pattern)
  • Dodson — interest-based nervous system / RSD
  • Hallowell & Ratey — ADHD 2.0
  • Mahan — the Wall of Awful

Built in public, with AI assistance (Claude).

About

An execution harness for ADHD: the intention->action coupling, not a motivation supply. Built one source at a time, in public.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages