create-dip

name	create-dip
description	Creates a Digital Identity Profile (DIP) for a new author from scratch. Collects raw sources (transcripts, voice, Telegram), extracts Tone of Voice, Stories Bank, Expertise Clusters. ONLY from raw sources, NEVER from posts.
disable-model-invocation	true
argument-hint	["Author-Name"]
allowed-tools	Read, Grep, Glob, Agent

Create Digital Identity Profile (DIP)

Build a full DIP for a new author from scratch.

Input

• $0 — the new author's name (e.g., "Author Name")

Required files

1. DIP Template: Templates/DIP-Template.md

2. DIP Creation Workflow: Templates/DIP-Creation-Workflow.md

3. Transcript map: From your repo's master CLAUDE.md → "Transcript map" section

⛔ CANONICAL SOURCE RULE

The DIP is built ONLY from raw sources:

• Transcripts of meetings, calls, webinars, podcasts
• The author's voice input
• Telegram posts personally written by the author
• CVs, articles, documents written by the author
• Transcripts of client interviews

NEVER from:

• LinkedIn posts (Posted/)
• PR comments
• Marketing copy
• Any AI-generated content

Why: DIP → Post → DIP = Feedback Loop. After 3–4 cycles the person's real voice is gone.

Pipeline

Phase 1: Source collection (30 min)

1. Identify the author's name
2. Look for transcripts across ALL locations on the transcript map:

Type	Location
Presales calls	Sales/Presales/Projects/Projects list/*/Transcripts/
Client meetings	Production/Projects/Projects list/*/Meetings transcripts/
AGI/Product calls	your-transcripts/internal-product-calls/
Granola (internal)	your-transcripts/synced-meetings/
Marketing calls	your-transcripts/marketing-meetings/
Podcasts/events	your-raw-sources/podcasts-and-events/
Webinars	your-transcripts/webinars/
Conferences	Media activities/Cursor Case Activities/Conferences/
Telegram	your-raw-sources/telegram-posts/

3. Minimum 3 transcripts, ideally 5+. Variety of types matters more than quantity.
4. If sources are thin → tell the user, request more material.

Phase 2: Tone of Voice extraction (60 min per transcript)

For each transcript, extract:

1. Characteristic phrases — how the person talks, their jargon
2. Reasoning style — how they build an argument
3. Speech patterns — start with an example? a thesis? a question?
4. Emotional register — serious? ironic? blunt? analytical?
5. What they NEVER say — "They NEVER" (forbidden patterns)

Phase 3: Build the Stories Bank (60 min)

From the transcripts, extract:

• Concrete cases with detail (client, situation, outcome)
• Personal stories (career moments, failures, learnings)
• Metrics and numbers from real experience
• Quotes — verbatim phrases that characterize the person

For each story, record:

• Source (file, approximate line)
• Context (who said it, when, to whom)
• Key details

Phase 4: Define Expertise Clusters (30 min)

Based on the transcripts, identify:

• 3–4 main expertise clusters
• Distribution across clusters (in %)
• Viral angles for each cluster

Phase 5: Fill the template

Load _DIP-Template.md and fill every section:

1. Identity & Position — who, role, company
2. Tone of Voice — style, speech patterns, "They NEVER"
3. Expertise Clusters — 3–4 clusters with distribution
4. Stories Bank — 10+ stories with sources
5. Viral Angles — formulas for content
6. Source Index — list of all sources used
7. Feedback Log — empty (filled as posts are published)

Phase 6: Save

Save to: Knowledge-Base/01-Identity-Profiles/[Author-Name]-DIP.md

Filename format: FirstName-LastName-DIP.md (hyphenated, title case)

Phase 7: Register

Add the author to:

1. Content/LinkedIn/CLAUDE.md (or your project's master routing file) → author table
2. .claude/CLAUDE.md → author DIPs section
3. Your master skill guide (if it has an author table)

Output

1. Finished DIP file — saved at the path above
2. Summary — brief recap: how many sources used, key author characteristics
3. Source Index — full list of files used
4. Recommendations — what else is needed for a full DIP (if sources are thin)

Versioning

Version	Minimum	Description
v0.5	1–2 sources	Bootstrap — basic tone, minimal stories
v0.7	3–4 sources	Working DIP — enough for posts, needs enrichment
v1.0	5+ sources, diverse types	Full DIP — ready for production