Outline format (talks-outline) — the depth bar

Every outline is a standalone .md the reader keeps and shares. Depth comes from granularity, not word count. A talk that enumerates many things becomes many ### subsections, never a table that swallows them. If a draft feels thin, split tables and paragraphs back into numbered subsections and per-line bullets — do not just add words.

Language

The whole file is in the language spoken in the talk (from meta.json language, verified against the transcript body; trust the body on conflict). Headers, table headers, tag legend, critical observations — all one language. Keep technical terms as spoken (DCF, ROE, TOD, pull request…). Only an explicit user override changes this; then note “translated from ” in the header.

Exhaustive-outline rules (apply to every talk)

• Preserve every anecdote, name (people/companies/tools), number, statistic, direct quote, and concrete example. Do not compress, generalize, or summarize away specifics.
• Deep decimal hierarchy: ## top sections (1, 2, 3…), ### subsections (1.1, 1.2…), #### where it helps. One ### per distinct sub-topic, example, or step. A long talk has 25+ ###, not 3.
• One idea per bullet line. Never pack several facts into one paragraph or table cell.
• Enumerations (patterns, steps, rules, trends, companies) -> each item its own numbered ### with its own bullets (rationale, metaphor, quote, example). A summary table at the end is a nice addition but must never replace the per-item subsections.
• Worked examples decomposed step by step — each micro-step its own ### (e.g. “7.1 Extract + placeholder name”, “7.2 Rename”).
• Q&A finely broken: one ### per question (Q1, Q2…; multi-part -> Q5, Q5a). Name the asker when known.
• Attribute quotes and claims to whoever said them. Quotes stay under 15 words.
• Open with a framing blockquote: speaker, venue, format, the talk’s own structure.
• End with a “People & References Cited” index.
• Enrich pass before writing: reread the transcript and add any anecdote, name, number, or example the outline missed.
• Shape matches the talk: a 90-min workshop -> very long outline; a narrative keynote or interview compresses to prose and is shorter at the same fidelity — expected, not a defect. Panels/podcasts -> outline as a topic-threaded conversation (still blockquote + References index). If a scheduled guest was absent or it became a roundtable, say so — never fabricate.

Analytical layer (this skill adds on top of the playbook)

Tag factual rows: [V] verifiable · [D] speaker’s estimate/opinion · [P] pending verification · [E] error or self-contradiction (state the correction). Speech-to-text garbles numbers and proper nouns — flag likely mis-transcriptions as [E]. Before the References index, add:

• Critical observations — analysis not in the talk: factual errors, contradictions, missing counter-arguments, sponsor/product incentives, sample-size or survivorship issues, claims stated as rules that rest on one anecdote.
• Pending verification — table of [P] claims with suggested primary sources.

Skeleton

---
title: "<Talk Title>"
date: "<YYYY-MM-DD>"
description: "<rich one-line description of what the outline covers>"
type: "reference"
tags: ["<domain>"]        # e.g. software / investing / self-help; add "ai" when apt
---

# <Talk Title>

> <Framing blockquote: speaker(s), venue/date, format (keynote/workshop/panel/podcast),
> and the talk's own structure. One paragraph.>

**Language:** <spoken language> · caption source: <manual | auto>
<add "translated captions only — treat wording with caution" or
 "outline translated from <language> at user request" when either applies>

## 1. <first section>
### 1.1 <sub-topic / example / step>
- one idea per bullet, tagged where factual  [V]/[D]/[P]/[E]
### 1.2 ...

## N. Q&A
### Q1 — <asker if known>: <question>
- <answer points>

## Critical observations (not in the talk)
| # | Issue | Why it matters to the reader |

## Pending verification
| Claim / figure | Suggested primary source |

## People & References Cited
- <person / company / tool / book / paper mentioned>

*Video: https://www.youtube.com/watch?v=<ID> — transcript fetched automatically; outline generated from the transcript.*

(For a long transcript captured across passes, append “(partial for a long transcript)” to the footer.)

Domain blocks (fold into the numbered sections where relevant)

Software engineering

Problem & context · architecture/approach (with trade-offs) · tools & versions (versions -> [P]) · reproducible workflow/steps · code/config shown (short snippets only) · perf / productivity claims (tagged, note if benchmarked) · where it holds vs breaks (monorepo vs micro-repo, solo vs team). Decompose every demo step by step.

Value investing / finance / real estate

Macro/cycle context (dated timeline) · business or asset (how it makes money, moat) · every figure in a tagged row with units/currency · valuation method + inputs (DCF/multiples/NAV) · policy/regulation with dates (usually [P]) · base/bull/bear outlook as stated + what was omitted · incentives (is the channel selling a course/fund/product?). Test the speaker’s thesis against their own evidence in Critical observations.

Self-help / productivity / mindset

Core model in one framework · evidence type per claim (study / anecdote / authority, tagged) · practices (how to do it, time cost, expected effect) · failure modes the speaker admits and ones they don’t · who it’s for / not for. Separate testable claims (a 2-week experiment) from motivational framing; flag pop-psych stated as settled science.

Style

Every row carries information; no filler. A 3-hour talk gets more sections, not longer rows. No more than one short (<15-word) verbatim quote per section. Never invent timestamps, figures, or names not in the transcript.