JoshuaShepherd Movement Leader Themes
Read a movement leader's own corpus and produce canonical theme documents in docs/themes/:
docs/themes/CORE_THEMES.md— taxonomy and decision record: which 4–6 themes earn pathway status, rejected candidates, integrated argument, book mapping.docs/themes/{slug}.md— one deep-dive per core theme: 12-section pathway content in the leader's voice, every substantive claim footnoted.
Discipline: themes are read out of the corpus, not imposed. Read full books before naming a theme. Research and config are check-against inputs, never source of truth.
Invocation
/movement-leader-themes {leader-slug}
/movement-leader-themes brad-brisco --taxonomy-only
/movement-leader-themes --theme discipleship
/movement-leader-themes --refresh
Run from the leader repo root.
| Flag | Effect |
|---|---|
{slug} or display name |
Defaults to cwd repo if obvious |
--taxonomy-only |
Write CORE_THEMES.md only |
--theme {slug} |
Rewrite one deep-dive (taxonomy must exist) |
--refresh |
Refresh content; keep existing slug set |
| (empty) | Infer leader from repo; ask if ambiguous |
Inputs (priority order)
| Source | Path | Weight |
|---|---|---|
| Book corpus | docs/books/{book-slug}/chapters/*.md |
Highest |
| Book metadata | docs/books/{book-slug}/*.md, README.md |
High |
| Research dossier | docs/movement_leader_research/{slug}/ |
Medium |
| Collated substrate | **/{SLUG}_RESEARCH_COLLATED.md |
Medium |
| Articles / resources | docs/**/articles/, docs/workspace/ |
Medium |
| Voice guide | docs/voice/{SLUG}_VOICE.md |
Required for deep-dives |
| Existing config | src/lib/config/tenant.config.ts, src/lib/content/pathways/*.ts |
Check-against only |
If docs/books/ is empty, stop and say so. Offer research-only run only if user confirms (lower confidence).
Execution protocol
Read this skill, then execute phases 1–4 to completion unless stopped by a flag.
Phase 1 — Inventory
ls docs/books/*/chapters/ 2>/dev/null
ls docs/movement_leader_research/ 2>/dev/null
ls docs/voice/ docs/themes/ 2>/dev/null
Build book list (slug, title, chapter count, year). For large corpora, delegate per-book reading via Task (subagent_type: explore, one agent per book → chapter-level theme notes), then synthesize.
Phase 2 — Taxonomy
Apply four-test filter (corpus centrality, distinctiveness, course viability, audience demand). Select 4–6 themes. Write docs/themes/CORE_THEMES.md per reference.md Part A. Stop if --taxonomy-only.
Phase 3 — Deep-dives
Ensure voice guide exists — if missing, read and run .cursor/skills/movement-leader-voice/SKILL.md first.
For each theme slug, write docs/themes/{slug}.md per reference.md Part B. Multi-theme runs: one Task agent per theme; each loads voice guide + relevant chapters.
Phase 4 — Verify
CORE_THEMES.md: all theme blocks complete; rejected-themes table present; decision record explains non-obvious calls- Each
{slug}.md: 12 sections; plural corpus-grounded cases; ≥2 distortions; ≥3 citations per substantive section; four necessities present - Cross-check: every taxonomy slug has a deep-dive (unless
--taxonomy-only); no orphan deep-dives
Print session report: slug, theme count, files written, voice guide used (y/n).
Key rules
- Read books first — no theme before its chapters are read.
- Leader's vocabulary — coined terms, not generic church-growth labels.
- Reject honestly — rejected-themes table and decision record mandatory.
- Every claim traces to corpus — footnotes with book + chapter + page (or
—). - Deep-dives in voice — load
docs/voice/{SLUG}_VOICE.md. - Cases plural and real — never invent contemporary examples.
- Confirm config by default — revise only when corpus contradicts; record why.
- Reference implementation: brad-brisco repo,
docs/themes/CORE_THEMES.mdanddocs/themes/{slug}.md.
Additional resources
- Taxonomy + deep-dive schemas: reference.md