ima-claude 2.18.0 → 2.25.0

package/plugins/ima-claude/skills/ima-copywriting/SKILL.md

@@ -7,226 +7,115 @@ metadata:
 
 # IMA Copywriting
 
-You are the editorial voice of the Independent Medical Alliance. Your goal is to write content that is clear, evidence-grounded, emotionally honest, and drives the reader to act — whether that's reading a study, attending a webinar, donating, or sharing.
-
-You are not a generic marketing copywriter. You write for a medical nonprofit that fights for independent medicine. Every piece of content reinforces IMA's position as a trustworthy, patient-centric alliance funded by people — not pharma.
-
----
-
-## Before Writing
-
-**Always load the `ima-brand` skill first.** That skill is the source of truth for voice, tone, terminology, and visual identity. This skill handles *editorial craft* — structure, rhythm, persuasion, and format-specific guidance.
-
----
+IMA = Independent Medical Alliance. Patient-centric, pharma-independent. Load `ima-brand` first (voice/tone authority).
 
 ## Content Type Router
 
-Identify the format, then read the corresponding reference file for template, structure, and voice notes.
-
-| Content Type | Reference File | Tone Blend |
-|-------------|----------------|------------|
-| 📬 Newsletter | [references/format-newsletter.md](references/format-newsletter.md) | Friendly + Informative |
-| 🎥 Webinar Email | [references/format-webinar-email.md](references/format-webinar-email.md) | Friendly + Informative (higher energy) |
-| 📝 Blog Post | [references/format-blog-post.md](references/format-blog-post.md) | Professional + Informative |
-| 📰 Press Release | [references/format-press-release.md](references/format-press-release.md) | Professional + Informative |
-| 💰 Fundraising Email | [references/format-fundraising-email.md](references/format-fundraising-email.md) | Inspirational + Supportive |
-| ✊ Op-Ed | [references/format-op-ed.md](references/format-op-ed.md) | Professional + Direct (earned edge) |
-| 📱 Social Post | [references/format-social-media.md](references/format-social-media.md) | Friendly + Informational |
-
-**Read only the reference file for the content type you're writing.** Do not load all formats.
-
----
-
-## IMA Writing Principles
+| Format | Reference File | Tone |
+|--------|----------------|------|
+| Newsletter | [references/format-newsletter.md](references/format-newsletter.md) | Friendly + Informative |
+| Webinar Email | [references/format-webinar-email.md](references/format-webinar-email.md) | Friendly + Informative (higher energy) |
+| Blog Post | [references/format-blog-post.md](references/format-blog-post.md) | Professional + Informative |
+| Press Release | [references/format-press-release.md](references/format-press-release.md) | Professional + Informative |
+| Fundraising Email | [references/format-fundraising-email.md](references/format-fundraising-email.md) | Inspirational + Supportive |
+| Op-Ed | [references/format-op-ed.md](references/format-op-ed.md) | Professional + Direct |
+| Social Post | [references/format-social-media.md](references/format-social-media.md) | Friendly + Informational |
 
-These principles override generic copywriting advice. They are specific to IMA's mission and audience.
+Read only the reference file for active content type.
 
-### 1. Evidence First, Emotion Second
-Lead with what the research shows. Earn emotional resonance through the weight of the evidence — don't manufacture it with hype.
+## Writing Principles
 
-❌ "This SHOCKING study changes EVERYTHING about cancer!"
-✅ "A major new review examines the relationship between COVID-19 vaccination and cancer — and the reaction to its publication tells its own story."
-
-### 2. Solution-Oriented, Never Fear-Based
-Acknowledge the problem clearly, then pivot to what IMA is doing about it. The reader should leave empowered, not paralyzed.
-
-❌ "Cancer rates are exploding and nobody is doing anything."
-✅ "Cancer is striking earlier, faster, and harder. At IMA, we're changing that."
-
-### 3. Independent, Not Adversarial
-IMA's independence is its strength. Name specific actors when relevant (NYT, AAP, pharmaceutical companies), but frame criticism through evidence and principle — not name-calling.
-
-❌ "Big Pharma shills at the NYT are lying again."
-✅ "The New York Times published a hit piece on Dr. Kirk Milhoan for holding a radical position: that patients and doctors should decide medical care together."
-
-### 4. Plain Language with Precision
-Write so a patient can understand and a physician won't cringe. Define medical terms on first use. Never dumb down the science — translate it.
-
-❌ "The study found bad stuff in the vaccines."
-✅ "New research explains why regulators keep missing DNA contamination in mRNA vaccines — and why independent labs keep finding it."
-
-### 5. The Reader Is an Ally, Not a Target
-IMA's audience chose to be here. Write to them as partners in a shared mission, not as prospects to convert. Respect their intelligence.
-
-❌ "You won't BELIEVE what we discovered..."
-✅ "Someone didn't want you to see this research."
-
-### 6. Honest Confidence
-State positions directly. Don't hedge when the evidence is strong. Don't overstate when it's emerging. The distinction is the brand.
-
-- Strong evidence: "Measles is highly treatable."
-- Emerging evidence: "Research points to mitochondrial dysfunction as a central driver."
-- Opinion: "We believe public health conversations should be rooted in data, not fear."
-
----
+1. **Evidence first** — Lead with research; earn emotion through weight of evidence, not hype.
+2. **Solution-oriented** — Acknowledge problem → pivot to IMA action. Reader leaves empowered, not paralyzed.
+3. **Independent, not adversarial** — Name actors when relevant; frame criticism through evidence, not name-calling.
+4. **Plain language with precision** — Patient understands, physician won't cringe. Define jargon on first use.
+5. **Reader is ally** — Partners in shared mission, not prospects to convert.
+6. **Honest confidence** — State strong evidence directly. Hedge emerging evidence. Label opinion.
 
-## Writing Style Rules
-
-### Voice Characteristics
+## Style Rules
 
 | Always | Never |
 |--------|-------|
-| Direct, declarative sentences | Clickbait or sensationalism |
+| Direct declarative sentences | Clickbait / sensationalism |
 | Active voice | Passive constructions |
 | Specific claims with evidence | Vague wellness-speak |
-| Conversational warmth | Stiff institutional tone |
-| Controlled intensity (when earned) | Breathless urgency on everything |
-| Humor and wit (when appropriate) | Snark without substance |
-
-### Word-Level Rules
-
-1. **Simple over complex** — "use" not "utilize," "help" not "facilitate"
-2. **Precise over vague** — Avoid "streamline," "optimize," "innovative," "cutting-edge"
-3. **Active over passive** — "IMA researchers discovered" not "It was discovered"
-4. **Earned intensity** — Phrases like "hit piece," "backfired," "the establishment is nervous" land because IMA uses them sparingly and backs them up
-5. **No exclamation points in body copy** — Exception: social media openers ("Live tonight!") and fundraising CTAs (one acceptable)
-6. **Define jargon** — "PACVS (Post-Acute COVID Vaccination Syndrome)" on first use
+| Earned intensity (used sparingly) | Breathless urgency |
+| "use" not "utilize" | "streamline," "optimize," "cutting-edge" |
 
-### IMA Terminology
+No exclamation points in body copy — exception: social openers, one fundraising CTA.
 
-| Avoid | Use Instead | Why |
-|-------|-------------|-----|
-| FLCCC | IMA / Independent Medical Alliance | Rebrand complete |
-| "Cure" / "treatment" (unqualified) | "Protocol" / "approach" / "may help with" | Precision |
-| "You should" | "You may consider" / "Research suggests" | Patient autonomy |
-| "Alternative medicine" | "Integrative medicine" / "complementary approaches" | Accurate framing |
-| "Big Pharma" (as sole descriptor) | Name the company or say "pharmaceutical industry" | Credibility |
-| "We believe" (for science) | "Evidence shows" / "Research indicates" | Evidence-based positioning |
-| "Anti-vaccine" (describing IMA) | "Pro-informed-consent" / "independent" | Accuracy |
+## Terminology
 
-### Transitions to Avoid (AI Tells)
+| Avoid | Use |
+|-------|-----|
+| FLCCC | IMA / Independent Medical Alliance |
+| "cure" / "treatment" (unqualified) | "protocol" / "approach" / "may help with" |
+| "you should" | "you may consider" / "research suggests" |
+| "alternative medicine" | "integrative medicine" / "complementary approaches" |
+| "Big Pharma" (sole descriptor) | company name or "pharmaceutical industry" |
+| "we believe" (for science) | "evidence shows" / "research indicates" |
+| "anti-vaccine" (describing IMA) | "pro-informed-consent" / "independent" |
 
-These phrases signal AI-generated content and undermine trust:
+## AI Tells — Never Use
 
-- "In today's rapidly evolving healthcare landscape..."
-- "It's worth noting that..."
-- "At its core,..."
-- "Let's delve into..."
-- "This begs the question..."
-- "In conclusion,..."
-- "Moreover,..." / "Furthermore,..." (overuse)
-
-**For approved transition phrases**: See [references/ima-transitions.md](references/ima-transitions.md)
-
----
+"In today's rapidly evolving landscape" · "It's worth noting" · "At its core" · "Let's delve into" · "In conclusion" · "Moreover" / "Furthermore" (overuse)
 
-## CTA Copy Guidelines
+See [references/ima-transitions.md](references/ima-transitions.md) for approved transitions.
 
-### CTA Patterns by Channel
+## CTA Patterns
 
-| Context | CTA Text | Notes |
-|---------|----------|-------|
-| Webinar / Blog (email/web) | "Watch Now »" / "Read More »" | Chevron (») is IMA email/web style |
-| Download (email/web) | "Download Now »" | For guides, PDFs |
-| Conference (email/web) | "🎟 Buy Tickets Now »" | Emoji before text |
-| Donate (primary) | "💙 Match My Gift Today »" | During match campaigns |
-| Donate (secondary) | "🌱 Help Us Grow Hope »" | Softer, end-of-email |
+| Context | CTA | Notes |
+|---------|-----|-------|
+| Webinar / Blog (email/web) | "Watch Now »" / "Read More »" | » required on email/web |
+| Download | "Download Now »" | |
+| Conference | "🎟 Buy Tickets Now »" | Emoji before text |
+| Donate (primary) | "💙 Match My Gift Today »" | Match campaigns |
+| Donate (secondary) | "🌱 Help Us Grow Hope »" | End-of-email |
 | Donate (evergreen) | "Please Support IMA's Mission »" | Standard newsletter close |
-| Social media | "Watch now." / "Learn more and sign:" | No chevron. Short imperative. |
+| Social | "Watch now." / "Learn more and sign:" | No chevron; short imperative |
 
-### CTA Rules
-1. **Email/web:** Always include » chevron at end
-2. **Social:** No chevron. Short imperative sentence ("Watch now.") or natural lead-in to link
-3. Emoji before CTA text on standalone email/web CTAs (💙, 🎟, 🌱, 👉)
-4. Verb-first: "Watch," "Read," "Download," "Support," "Buy"
-5. Never: "Click Here," "Learn More," "Submit"
-6. One primary CTA per content section; repeat in P.S. for fundraising
-
----
+Verb-first. One primary CTA per section; repeat in P.S. for fundraising. Never "Click Here," "Learn More," "Submit."
 
 ## Structural Patterns
 
-### The "Honest Medicine" Close (Newsletters)
-
-Every newsletter ends with this block. Vary the content; keep the structure:
-
+**"Honest Medicine" Close** (every newsletter):
 ```
 PLEASE SUPPORT HONEST MEDICINE
 
-[2-3 sentences connecting this week's specific outputs to what donors fund.]
-No pharma funding, no institutional strings. Just [description of what IMA does].
+[2-3 sentences: this week's outputs → what donors fund.]
+No pharma funding, no institutional strings. Just [what IMA does].
 
 If you've found value in what we do, your support helps us keep going.
 
 Please Support IMA's Mission »
 ```
 
-### The Opening Hook (Newsletters)
+**Opening Hook Patterns:**
 
 | Pattern | Example |
 |---------|---------|
 | Bold declaration | "The establishment is nervous." |
 | Provocative setup | "Someone didn't want you to see this research." |
-| Trend observation | "Peptides are popping up like no tomorrow in the world of wellness." |
+| Trend | "Peptides are popping up like no tomorrow in wellness." |
 | Contrast/irony | "Apparently, this is now a radical position." |
 
-### The Quick Links Block (Newsletters)
-
-Always emoji + title format. 3-6 items:
-
-| Emoji | Use For |
-|-------|---------|
-| 💊 | Drug/therapy/peptide content |
-| 📰 | News, responses, op-eds |
-| 🔬 | Research, studies, journal articles |
-| 🧬 | Genetics, DNA, molecular biology |
-| ⚡ | Breaking/urgent content |
-| 💬 | Opinion, commentary, "Here's a Thought" |
-| 💪 | Health guides, wellness content |
-| 💤 | Sleep-related content |
-| 🎉 | Achievements, appointments, wins |
-| 📖 | Journal, publications, long reads |
-| 📄 | Guides, downloadables, PDFs |
-
----
+**Quick Links Block** (newsletters, emoji + title, 3-6 items):
+💊 therapy · 📰 news/op-eds · 🔬 research · 🧬 genetics · ⚡ breaking · 💬 opinion · 💪 wellness · 💤 sleep · 🎉 wins · 📖 publications · 📄 guides
 
 ## Quality Checklist
 
-Before delivering any draft, verify:
-
-- [ ] **Brand alignment**: Does this sound like IMA, not a generic health org?
-- [ ] **Evidence backing**: Is every health claim supported or clearly labeled?
-- [ ] **Terminology**: No "FLCCC," no "cure" without context, no loaded terms?
-- [ ] **Active voice**: < 10% passive constructions?
-- [ ] **Plain language**: Would a patient understand this? Would a physician respect it?
-- [ ] **CTA clarity**: Does the reader know exactly what to do next?
-- [ ] **CTA format**: Chevron for email/web, short imperative for social?
-- [ ] **Emotional honesty**: Urgency earned, not manufactured?
-- [ ] **AI tells**: No "delve," "landscape," "moreover" chains, "it's worth noting"?
-- [ ] **Tone match**: Does the tone fit the format and channel?
-- [ ] **Independence signal**: Does the reader understand IMA is not pharma-funded?
-
----
-
-## Additional References
-
-- **[references/ima-transitions.md](references/ima-transitions.md)** — Approved transition phrases
-- **[references/ima-copy-frameworks.md](references/ima-copy-frameworks.md)** — Headline formulas, structural templates, section patterns
-
----
-
-## Related Skills
-
-- **ima-brand**: Voice, tone, terminology, visual identity (ALWAYS load alongside this skill)
-- **ima-editorial-scorecard**: Score and assess existing content against IMA standards
-- **ima-bootstrap**: CSS/SCSS for web implementation of IMA content
+- [ ] Sounds like IMA, not generic health org?
+- [ ] Every health claim supported or labeled?
+- [ ] No "FLCCC," no unqualified "cure"?
+- [ ] < 10% passive voice?
+- [ ] Patient understands? Physician respects?
+- [ ] Clear next action for reader?
+- [ ] Chevron on email/web CTAs; short imperative on social?
+- [ ] Urgency earned, not manufactured?
+- [ ] No AI tells?
+- [ ] IMA independence signal present?
+
+## References
+
+- [references/ima-transitions.md](references/ima-transitions.md) — Approved transition phrases
+- [references/ima-copy-frameworks.md](references/ima-copy-frameworks.md) — Headline formulas, structural templates

package/plugins/ima-claude/skills/ima-doc2pdf/SKILL.md

@@ -14,37 +14,19 @@ description: >-
 
 # IMA DOCX → Branded PDF
 
-Extracts content from any Word document and generates a branded IMA PDF with
-Lato typography, navy/gold colors, and IMA layout standards. Works with guides,
-reports, white papers, and other IMA documents. Outputs content pages only
-(no cover page) — the cover is handled by `ima-cover-creator` and merged via pypdf.
-
----
-
-## Why ReportLab for Content
-
-Content pages are pure document flow: headings, body paragraphs, bullet lists,
-inline bold/italic, embedded images, and running footers. ReportLab handles all of
-this with proper Lato font registration, automatic page breaks, and precise
-typographic control. No coordinate math needed — the flow engine does the work.
-
----
+Extracts content from Word documents, generates branded IMA PDF (Lato, navy/gold, IMA layout). Outputs content pages only — cover handled by `ima-cover-creator` and merged via pypdf.
 
 ## Quick Start
 
-### 1. Install dependencies (once)
-
 ```bash
+# Install (once)
 pip install python-docx reportlab Pillow pypdf --break-system-packages
-```
 
-### 2. Generate content PDF
-
-```bash
+# Generate content PDF
 python3 scripts/generate_pdf.py path/to/document.docx --out content.pdf
 ```
 
-### 3. Merge with cover (from ima-cover-creator)
+## Merge with Cover
 
 ```python
 from pypdf import PdfReader, PdfWriter
@@ -55,7 +37,7 @@ content = PdfReader("content.pdf")
 writer = PdfWriter()
 writer.add_page(cover.pages[0])
 
-# Skip the first 2 pages (ReportLab's placeholder cover + overflow)
+# Skip first 2 pages (ReportLab placeholder cover + overflow)
 for page in content.pages[2:]:
     writer.add_page(page)
 
@@ -63,28 +45,19 @@ with open("final.pdf", "wb") as f:
     writer.write(f)
 ```
 
----
-
 ## Pipeline
 
 ```
-DOCX
-  ↓  extract_docx.py (text, structure, metadata)
-  ↓  generate_pdf.py (ReportLab → branded PDF)
-  ↓
-content.pdf (N pages, no cover)
-  +
-cover.pdf (from ima-cover-creator)
-  ↓  pypdf merge
-  ↓
-final.pdf → Canva import
+DOCX → extract_docx.py → generate_pdf.py → content.pdf
+                                                +
+                                           cover.pdf (ima-cover-creator)
+                                                ↓ pypdf merge
+                                           final.pdf → Canva import
 ```
 
----
+## Typography Spec
 
-## Typography Spec (Canva-confirmed)
-
-All values confirmed from Canva design data. Font: Lato (Google Fonts).
+Font: Lato (Google Fonts). Values confirmed from Canva design data.
 
 ### Headings
 
@@ -124,18 +97,16 @@ All values confirmed from Canva design data. Font: Lato (Google Fonts).
 | Body width | 7.5 in |
 | Footer height | 0.4 in from bottom |
 
----
-
 ## Content Extraction
 
-The `extract_docx.py` script classifies each Word paragraph into typed blocks:
+`extract_docx.py` classifies each paragraph into typed blocks. Each block includes `runs` with per-run bold/italic flags.
 
 | Type | Description |
 |------|-------------|
-| `h1` | Top-level heading (title) |
+| `h1` | Top-level heading |
 | `h2` | Section heading |
 | `h3` | Sub-heading |
-| `heading_bold` | All-bold paragraph (inline heading) |
+| `heading_bold` | All-bold paragraph |
 | `body` | Regular paragraph |
 | `bullet` | List item |
 | `author` | Author name |
@@ -146,97 +117,56 @@ The `extract_docx.py` script classifies each Word paragraph into typed blocks:
 | `answer_start` | Q&A answer (YES/NO prefix) |
 | `reference` | Numbered citation |
 | `ref_heading` | "References" heading |
-| `figure_caption` | Figure/table caption |
+| `figure_caption` | Caption |
 | `page_break` | Hard page break |
 
-Each block includes `runs` with per-run bold/italic flags for inline formatting.
-
----
-
 ## Image Handling
 
-The script extracts embedded DOCX images via `python-docx`:
-
-1. Reads all image relationships from the DOCX package
-2. Maps paragraph indices to embedded image positions
-3. Writes images to temp files
-4. Inserts ReportLab `Image` flowables at the correct positions
-5. Scales to fit within `page_width - 2 × margin`
-
-Images that appear between text paragraphs (image-only paragraphs) are also caught
-and appended after all text content.
-
----
+1. Read all image relationships from DOCX package
+2. Map paragraph indices to embedded image positions
+3. Write images to temp files
+4. Insert ReportLab `Image` flowables at correct positions
+5. Scale to fit `page_width - 2 × margin`
 
 ## Cover Page Behavior
 
-The ReportLab script generates a **placeholder cover** (navy background with title
-text) as pages 1-2 of its output. This exists so the script works standalone, but
-when pairing with `ima-cover-creator`, **skip the first 2 pages** during merge.
+ReportLab generates placeholder cover as pages 1-2. Skip them during merge. To verify:
 
-To check which pages to skip:
 ```python
 from pypdf import PdfReader
 r = PdfReader("content.pdf")
 for i in range(min(3, len(r.pages))):
-    text = r.pages[i].extract_text()[:100]
-    print(f"Page {i}: {text}")
+    print(f"Page {i}: {r.pages[i].extract_text()[:100]}")
 ```
 
-The first content page typically starts with "Introduction" or a section heading.
-
----
+First content page typically starts with "Introduction" or section heading.
 
 ## Scripts
 
 | Script | Purpose |
 |--------|---------|
-| `generate_pdf.py` | Main: DOCX → branded PDF via ReportLab |
+| `generate_pdf.py` | DOCX → branded PDF via ReportLab |
 | `extract_docx.py` | Extracts structured content from Word documents |
-| `docx_utils.py` | Shared utilities for DOCX parsing |
-
----
+| `docx_utils.py` | Shared DOCX parsing utilities |
 
 ## Fonts
 
-Lato TTF files are auto-downloaded from Google Fonts on first run into the `fonts/`
-directory (which is git-ignored). The font family (Regular, Bold, Italic, BoldItalic)
-is registered with ReportLab so that `<b>` and `<i>` markup works in Paragraph objects.
-
-If the fonts are already present, the download is skipped.
-
----
+Lato TTF auto-downloaded from Google Fonts on first run into `fonts/` (git-ignored). Family (Regular, Bold, Italic, BoldItalic) registered with ReportLab. Download skipped if fonts present.
 
 ## Customization
 
-### Adjusting the footer
-
-The footer shows the document title and date. To customize:
 ```python
-# In generate_pdf.py, the footer text is built from:
+# Footer text (in generate_pdf.py)
 title_short = (title[:60] + "...") if len(title) > 60 else title
 footer_text = f"{title_short} ({clean_date})"
 ```
 
-### Adding new paragraph types
+To add paragraph types: add classifier rule in `extract_docx.py`, handler in `block_to_flowables()`, and `ParagraphStyle` in `build_styles()`.
 
-1. Add a classifier rule in `extract_docx.py`
-2. Add a handler in `block_to_flowables()` in `generate_pdf.py`
-3. Create a ReportLab `ParagraphStyle` in `build_styles()`
-
----
-
-## Relationship to Other Skills
+## Related Skills
 
 | Skill | Role |
 |-------|------|
-| **ima-cover-creator** | Generates branded cover page (PPTX → PDF) |
-| **ima-cancer-care-guides** | Full pipeline including Markdown source and Canva API mapping |
+| **ima-cover-creator** | Branded cover page (PPTX → PDF) |
+| **ima-cancer-care-guides** | Full pipeline with Markdown source and Canva API |
 | **ima-brand** | Source of truth for colors, typography, voice |
-
-**Typical workflow:**
-```
-ima-cover-creator  →  cover.pdf (1 page)
-ima-doc2pdf        →  content.pdf (skip first 2 pages)
-pypdf merge        →  final.pdf → Canva import
-```