Produce a fragmentation story for: $ARGUMENTS
$ARGUMENTS is one or more movement leaders — either the folder slug
(alan-hirsch, andy-crouch) or a full name (Alan Hirsch). Multiple leaders
may be passed; handle each independently and write a separate story for each.
What This Skill Produces
For each named leader, a single markdown file at
docs/movement_leader_research/{slug}/fragmentation-story.md containing:
- The narrative opening — 2–4 paragraphs telling their specific story of
informational and relational fragmentation. Not a template. Names, specific
counts, specific places. The reader should feel the scatter.
- Informational fragmentation map — an exhaustive inventory of every
surface where their content lives, with titles and locations filled in
to the level the research supports.
- Relational fragmentation map — where their relationships, networks, and
collaborations live (and don't live).
- The synthesis — a short closing that makes the picture legible: what
fraction of their work is retrievable, what is trapped, what is invisible.
This is not generic. Blank fields must be labeled UNKNOWN — needs research
so the gaps are as honest as the findings.
Step 1 — Resolve the leader
- Normalize the argument. If a full name was passed, slugify it
(
Alan Hirsch → alan-hirsch). If passed a slug, use it directly.
- Check
docs/movement_leader_research/{slug}/ exists. If not, also check
docs/movement_leader_research/profiles/{slug}/. If neither exists, stop
and tell the user no research folder was found — point them at
/author-research to create one.
- Note the folder path you resolved for the rest of the run.
Step 2 — Inventory the available research
Read, in order of priority, whichever of these exist inside the resolved folder:
sources.md — raw source list, often with URLs
profile/ — identity.md, biography.md, theology.md,
voice-analysis.md, calling-profile.md
content/books.md, content/articles.md, content/academic.md,
content/audio.md, content/videos.md, content/courses.md,
content/frameworks.md, content/content-audit.md
digital-presence/websites.md, digital-presence/social-media.md,
digital-presence/newsletters.md, digital-presence/platforms.md
summary.md, network/organizations.md, analysis/audience-analysis.md
_misc/ helpers — content-analysis-root.md,
digital-presence-discovery.md, network/ (collaborators, endorsements,
events), media/ (citations, press-coverage, reviews)
_staff/ helpers — gap-analysis.md, plus the older legacy/ ALL-CAPS
profiles ({NAME}_CONTENT_AUDIT.md, {NAME}_AUTHOR_PROFILE.md,
{NAME}_ORGS.md, {NAME}_TIMELINE.md) only if present — a freshly
researched leader has the categorized files above instead
- For Alan Hirsch specifically, also read
docs/movement_leader_research/alan-hirsch-baseline-report.md — it has
high-confidence counts and constraint analysis.
Use Glob (docs/movement_leader_research/{slug}/**/*.md) to see the full tree,
then Read each relevant file. Do not guess at content that is not in the
research. Mark gaps honestly.
Step 3 — Build the informational inventory
For every category below, list specific titles with specific locations
(URL, filesystem path, publisher, platform name). If a category has no
evidence in the research, write UNKNOWN — needs research rather than
omitting it.
Required categories (include all, even when empty):
- Books — title, publisher, year, co-authors, where it's actually
sold/distributed (Amazon, IVP, publisher site), audiobook status
- Academic / peer-reviewed work — dissertations, chapters, journal
articles, with the journal/publisher name
- PDFs & downloadable documents — whitepapers, Lausanne papers, study
guides; where they're hosted (if known)
- Articles / essays — platform + title + year; note any that exist only
as links that may be dead
- Blogs / personal writing — every blog surface they've used
(Substack, Medium, personal site, guest posts on Forge/Saturate/etc.)
- Websites they control or co-own — personal domain, org sites, legacy
project sites
- Social media — each account with the handle, platform, approximate
follower count if known, posting frequency if known
- Newsletters / email — platform (Substack, Mailchimp, ConvertKit), list
size if known, cadence
- Podcasts (hosted) — shows they own/host, where they publish
- Podcasts (guest appearances) — every named episode/show from the
research
- YouTube / video — own channel (or absence of one), videos on other
channels by name
- Sermons / teaching audio — church archives, conference recordings,
specific series if named
- Talks / conference presentations — conference name, year, topic if
known
- Courses — seminary, online platform, self-hosted; platform name and
institution
- Assessments / tools — APEST, diagnostic instruments, apps they've
built; where they live
- Apps / software platforms — any digital products
- Notes / laptop / unpublished — explicitly label this
UNKNOWN — likely substantial, not yet catalogued unless the research
says otherwise
- Their own head / embodied knowledge — frameworks that exist primarily
through speaking, consulting, coaching; 1-on-1 wisdom not yet externalized
For each item where a URL exists in the research, include the URL. For each
that is hinted at but unverified, add (unverified).
Step 4 — Build the relational inventory
- Organizations — founded, board roles, advisory relationships (with
org names)
- Co-authors and frequent collaborators — names, which works they
co-created
- Networks / movements — the communities they are nodes in
(Forge, Exponential, Lausanne, 100M, etc.)
- Seminary / academic affiliations — institutions, role titles
- Mentors and protégés — if the research names them
- Where relationships are stored — CRM? Personal address book? Mental
rolodex? Email history? Mark honestly — usually this is the most
fragmented layer.
Step 5 — Write the narrative
The file must open with prose that tells this person's story, not a
template. Use the specific numbers the research supplies. Good opens look
like:
Alan Hirsch has written 20 books across 7 publishers over 21 years. More
than 150,000 people have taken his APEST assessment. And if you try to
follow his thinking from first book to current frame, you cannot — because
there is no path. The books live on Amazon. The assessment lives on
5qcentral.com with no onward journey. His articles are scattered across
Saturate, Verge, Exponential, and a static Squarespace at alanhirsch.org.
His 38+ videos live on 20+ third-party YouTube channels. He does not have
his own...
Bad opens look like: "Leader X has a significant body of work spread across
multiple channels." Do not write that.
Follow the opening with the informational map, the relational map, and a
one-paragraph synthesis that states explicitly: how retrievable their
knowledge is, what fraction is embodied vs. externalized, what the single
largest pool of "trapped" value is.
Step 6 — Output
Write the file to
docs/movement_leader_research/{slug}/fragmentation-story.md.
Use this heading structure exactly:
# {Full Name}: The Fragmentation Story
> Generated {YYYY-MM-DD} by the `fragmentation-story` skill from existing
> research in `docs/movement_leader_research/{slug}/`.
## The Story
{2–4 paragraphs of specific narrative}
## Informational Fragmentation
### Books
### Academic Work
### PDFs & Documents
### Articles
### Blogs & Personal Writing
### Websites
### Social Media
### Newsletters / Email
### Podcasts (Hosted)
### Podcasts (Guest)
### YouTube & Video
### Sermons / Teaching Audio
### Talks / Conferences
### Courses
### Assessments & Tools
### Apps / Software
### Notes, Laptop, Unpublished
### Embodied / In Their Head
## Relational Fragmentation
### Organizations
### Co-authors & Collaborators
### Networks & Movements
### Academic Affiliations
### Mentors, Protégés, Key Relationships
### Where Relationships Are Actually Stored
## Synthesis
{one paragraph}
## Research Provenance
- Source folder: `docs/movement_leader_research/{slug}/`
- Files read: {list the files you actually read}
- Confidence: {one sentence — what's well-evidenced vs. what's thin}
- Gaps requiring further research: {bullet list of the biggest UNKNOWNs}
If the user passed multiple leaders, repeat Step 2 through Step 6 for each,
then print a short summary to the user listing each file you wrote.
Quality Bar
- Specific, not generic. Titles, URLs, platforms, handles. If the research
has them, the story has them.
- Honest gaps.
UNKNOWN — needs research beats a plausible invention.
- No downstream recommendations. This skill diagnoses fragmentation; it
does not prescribe the Movemental platform response. That's a different
document.
- Do not recompute from scratch. The point is to organize what's
already been researched, not to re-scrape. If a category is genuinely
empty after reading everything, say so.
1---2name: fragmentation-story3description: >4 Tell a named movement leader's actual story of informational and relational5 fragmentation. Inventories every place their content currently lives — books,6 PDFs, articles, blogs, websites, social media, their own head, notes/laptop,7 apps, podcasts, YouTube, sermons, talks — with specific titles and locations8 pulled from prior research. Use when asked to map fragmentation for a leader,9 audit where their content is scattered, or produce a "where does it all live"10 report. Reads from `docs/movement_leader_research/{slug}/` and writes11 `fragmentation-story.md` into that leader's folder.12disable-model-invocation: true13metadata:14 source: https://skillmd.com/skills/joshuashepherd/fragmentation-story15---1617Produce a fragmentation story for: $ARGUMENTS1819`$ARGUMENTS` is one or more movement leaders — either the folder slug20(`alan-hirsch`, `andy-crouch`) or a full name (`Alan Hirsch`). Multiple leaders21may be passed; handle each independently and write a separate story for each.2223## What This Skill Produces2425For each named leader, a single markdown file at26`docs/movement_leader_research/{slug}/fragmentation-story.md` containing:27281. **The narrative opening** — 2–4 paragraphs telling *their specific story* of29 informational and relational fragmentation. Not a template. Names, specific30 counts, specific places. The reader should feel the scatter.312. **Informational fragmentation map** — an exhaustive inventory of every32 surface where their content lives, with **titles and locations filled in**33 to the level the research supports.343. **Relational fragmentation map** — where their relationships, networks, and35 collaborations live (and don't live).364. **The synthesis** — a short closing that makes the picture legible: what37 fraction of their work is retrievable, what is trapped, what is invisible.3839This is not generic. Blank fields must be labeled **UNKNOWN — needs research**40so the gaps are as honest as the findings.4142## Step 1 — Resolve the leader43441. Normalize the argument. If a full name was passed, slugify it45 (`Alan Hirsch` → `alan-hirsch`). If passed a slug, use it directly.462. Check `docs/movement_leader_research/{slug}/` exists. If not, also check47 `docs/movement_leader_research/profiles/{slug}/`. If neither exists, stop48 and tell the user no research folder was found — point them at49 `/author-research` to create one.503. Note the folder path you resolved for the rest of the run.5152## Step 2 — Inventory the available research5354Read, in order of priority, whichever of these exist inside the resolved folder:5556- `sources.md` — raw source list, often with URLs57- `profile/` — `identity.md`, `biography.md`, `theology.md`,58 `voice-analysis.md`, `calling-profile.md`59- `content/books.md`, `content/articles.md`, `content/academic.md`,60 `content/audio.md`, `content/videos.md`, `content/courses.md`,61 `content/frameworks.md`, `content/content-audit.md`62- `digital-presence/websites.md`, `digital-presence/social-media.md`,63 `digital-presence/newsletters.md`, `digital-presence/platforms.md`64- `summary.md`, `network/organizations.md`, `analysis/audience-analysis.md`65- `_misc/` helpers — `content-analysis-root.md`,66 `digital-presence-discovery.md`, `network/` (collaborators, endorsements,67 events), `media/` (citations, press-coverage, reviews)68- `_staff/` helpers — `gap-analysis.md`, plus the older `legacy/` ALL-CAPS69 profiles (`{NAME}_CONTENT_AUDIT.md`, `{NAME}_AUTHOR_PROFILE.md`,70 `{NAME}_ORGS.md`, `{NAME}_TIMELINE.md`) only if present — a freshly71 researched leader has the categorized files above instead72- For Alan Hirsch specifically, also read73 `docs/movement_leader_research/alan-hirsch-baseline-report.md` — it has74 high-confidence counts and constraint analysis.7576Use Glob (`docs/movement_leader_research/{slug}/**/*.md`) to see the full tree,77then Read each relevant file. Do not guess at content that is not in the78research. Mark gaps honestly.7980## Step 3 — Build the informational inventory8182For every category below, list **specific titles with specific locations**83(URL, filesystem path, publisher, platform name). If a category has no84evidence in the research, write `UNKNOWN — needs research` rather than85omitting it.8687Required categories (include all, even when empty):88891. **Books** — title, publisher, year, co-authors, where it's actually90 sold/distributed (Amazon, IVP, publisher site), audiobook status912. **Academic / peer-reviewed work** — dissertations, chapters, journal92 articles, with the journal/publisher name933. **PDFs & downloadable documents** — whitepapers, Lausanne papers, study94 guides; where they're hosted (if known)954. **Articles / essays** — platform + title + year; note any that exist only96 as links that may be dead975. **Blogs / personal writing** — every blog surface they've used98 (Substack, Medium, personal site, guest posts on Forge/Saturate/etc.)996. **Websites they control or co-own** — personal domain, org sites, legacy100 project sites1017. **Social media** — each account with the handle, platform, approximate102 follower count if known, posting frequency if known1038. **Newsletters / email** — platform (Substack, Mailchimp, ConvertKit), list104 size if known, cadence1059. **Podcasts (hosted)** — shows they own/host, where they publish10610. **Podcasts (guest appearances)** — every named episode/show from the107 research10811. **YouTube / video** — own channel (or absence of one), videos on other109 channels by name11012. **Sermons / teaching audio** — church archives, conference recordings,111 specific series if named11213. **Talks / conference presentations** — conference name, year, topic if113 known11414. **Courses** — seminary, online platform, self-hosted; platform name and115 institution11615. **Assessments / tools** — APEST, diagnostic instruments, apps they've117 built; where they live11816. **Apps / software platforms** — any digital products11917. **Notes / laptop / unpublished** — explicitly label this120 `UNKNOWN — likely substantial, not yet catalogued` unless the research121 says otherwise12218. **Their own head / embodied knowledge** — frameworks that exist primarily123 through speaking, consulting, coaching; 1-on-1 wisdom not yet externalized124125For each item where a URL exists in the research, include the URL. For each126that is hinted at but unverified, add `(unverified)`.127128## Step 4 — Build the relational inventory1291301. **Organizations** — founded, board roles, advisory relationships (with131 org names)1322. **Co-authors and frequent collaborators** — names, which works they133 co-created1343. **Networks / movements** — the communities they are nodes in135 (Forge, Exponential, Lausanne, 100M, etc.)1364. **Seminary / academic affiliations** — institutions, role titles1375. **Mentors and protégés** — if the research names them1386. **Where relationships are stored** — CRM? Personal address book? Mental139 rolodex? Email history? Mark honestly — usually this is the most140 fragmented layer.141142## Step 5 — Write the narrative143144The file must open with prose that tells *this person's* story, not a145template. Use the specific numbers the research supplies. Good opens look146like:147148> Alan Hirsch has written 20 books across 7 publishers over 21 years. More149> than 150,000 people have taken his APEST assessment. And if you try to150> follow his thinking from first book to current frame, you cannot — because151> there is no path. The books live on Amazon. The assessment lives on152> 5qcentral.com with no onward journey. His articles are scattered across153> Saturate, Verge, Exponential, and a static Squarespace at alanhirsch.org.154> His 38+ videos live on 20+ third-party YouTube channels. He does not have155> his own...156157Bad opens look like: "Leader X has a significant body of work spread across158multiple channels." Do not write that.159160Follow the opening with the informational map, the relational map, and a161one-paragraph synthesis that states explicitly: how retrievable their162knowledge is, what fraction is embodied vs. externalized, what the single163largest pool of "trapped" value is.164165## Step 6 — Output166167Write the file to168`docs/movement_leader_research/{slug}/fragmentation-story.md`.169170Use this heading structure exactly:171172```markdown173# {Full Name}: The Fragmentation Story174175> Generated {YYYY-MM-DD} by the `fragmentation-story` skill from existing176> research in `docs/movement_leader_research/{slug}/`.177178## The Story179180{2–4 paragraphs of specific narrative}181182## Informational Fragmentation183184### Books185### Academic Work186### PDFs & Documents187### Articles188### Blogs & Personal Writing189### Websites190### Social Media191### Newsletters / Email192### Podcasts (Hosted)193### Podcasts (Guest)194### YouTube & Video195### Sermons / Teaching Audio196### Talks / Conferences197### Courses198### Assessments & Tools199### Apps / Software200### Notes, Laptop, Unpublished201### Embodied / In Their Head202203## Relational Fragmentation204205### Organizations206### Co-authors & Collaborators207### Networks & Movements208### Academic Affiliations209### Mentors, Protégés, Key Relationships210### Where Relationships Are Actually Stored211212## Synthesis213214{one paragraph}215216## Research Provenance217218- Source folder: `docs/movement_leader_research/{slug}/`219- Files read: {list the files you actually read}220- Confidence: {one sentence — what's well-evidenced vs. what's thin}221- Gaps requiring further research: {bullet list of the biggest UNKNOWNs}222```223224If the user passed multiple leaders, repeat Step 2 through Step 6 for each,225then print a short summary to the user listing each file you wrote.226227## Quality Bar228229- **Specific, not generic.** Titles, URLs, platforms, handles. If the research230 has them, the story has them.231- **Honest gaps.** `UNKNOWN — needs research` beats a plausible invention.232- **No downstream recommendations.** This skill diagnoses fragmentation; it233 does not prescribe the Movemental platform response. That's a different234 document.235- **Do not recompute from scratch.** The point is to *organize* what's236 already been researched, not to re-scrape. If a category is genuinely237 empty after reading everything, say so.