ima-claude 2.18.0 → 2.20.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +7 -0
- package/dist/cli.js +5 -1
- package/package.json +1 -1
- package/plugins/ima-claude/.claude-plugin/plugin.json +2 -2
- package/plugins/ima-claude/skills/agentic-workflows/SKILL.md +133 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/phases/deliver.md +181 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/phases/draft.md +99 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/phases/gather.md +130 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/phases/outline.md +106 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/phases/review.md +137 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/standards/draft-format.md +159 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/standards/editorial-standards.md +160 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/standards/outline-format.md +110 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/avada-construction-guide.md +263 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/avada-webinar-example.txt +275 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/cta-block-catalog.md +169 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/espo-email-preparation.md +241 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/webinar-recap-email-espo.html +339 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/templates/webinar-reminder-email-espo.html +458 -0
- package/plugins/ima-claude/skills/agentic-workflows/references/workflows/editorial/webinar-summary.md +81 -0
- package/plugins/ima-claude/skills/design-to-code/SKILL.md +126 -0
- package/plugins/ima-claude/skills/design-to-code/references/guardrails.md +46 -0
- package/plugins/ima-claude/skills/design-to-code/references/phase-a-design-to-prompt.md +141 -0
- package/plugins/ima-claude/skills/design-to-code/references/phase-b-prompt-to-code.md +155 -0
- package/plugins/ima-claude/skills/design-to-code/references/prompt-template.md +95 -0
- package/plugins/ima-claude/skills/mcp-atlassian/SKILL.md +94 -14
- package/plugins/ima-claude/skills/mcp-atlassian/references/direct-api-attachments.md +115 -0
- package/plugins/ima-claude/skills/mcp-atlassian/references/direct-api-auth.md +103 -0
- package/plugins/ima-claude/skills/mcp-atlassian/references/direct-api-bulk.md +149 -0
- package/plugins/ima-claude/skills/mcp-atlassian/references/direct-api-misc.md +195 -0
- package/plugins/ima-claude/skills/mcp-atlassian/references/direct-api-sprints.md +158 -0
- package/plugins/ima-claude/skills/prompt-starter/SKILL.md +9 -6
|
@@ -0,0 +1,106 @@
|
|
|
1
|
+
# Phase 2: Outline
|
|
2
|
+
|
|
3
|
+
You are the structural editor for a headless content pipeline. Your job is to propose the content architecture that best serves the reader — not the one that mirrors the source material's sequence.
|
|
4
|
+
|
|
5
|
+
This is a single-pass operation. Produce a complete outline. Do not defer decisions.
|
|
6
|
+
|
|
7
|
+
---
|
|
8
|
+
|
|
9
|
+
## Input Format
|
|
10
|
+
|
|
11
|
+
The consumer will provide:
|
|
12
|
+
|
|
13
|
+
- **Gather phase output** — structured summary with speakers, scope, word target, and assets
|
|
14
|
+
- **Recipe overrides** — content-type-specific outline conventions (appended by consumer, if any)
|
|
15
|
+
- **Standards** — the editorial standards and outline format rules injected by the consumer
|
|
16
|
+
|
|
17
|
+
---
|
|
18
|
+
|
|
19
|
+
## Process
|
|
20
|
+
|
|
21
|
+
### 1. Read the Gather Output
|
|
22
|
+
|
|
23
|
+
Extract the key elements you need:
|
|
24
|
+
- Core subject matter and main findings
|
|
25
|
+
- Word target (carry it forward)
|
|
26
|
+
- Available assets
|
|
27
|
+
- Speakers and their roles
|
|
28
|
+
|
|
29
|
+
### 2. Identify the Core Argument
|
|
30
|
+
|
|
31
|
+
Every piece of content has one central claim or insight the reader should leave with. State it to yourself before building the outline. Every section must serve this argument.
|
|
32
|
+
|
|
33
|
+
### 3. Structure Around the Reader
|
|
34
|
+
|
|
35
|
+
Build the outline in the order that serves a reader who is encountering this material for the first time. Do not follow the source material's sequence unless it happens to be the best reader sequence.
|
|
36
|
+
|
|
37
|
+
Apply the outline format rules provided in the injected standards.
|
|
38
|
+
|
|
39
|
+
### 4. Place Visual Elements
|
|
40
|
+
|
|
41
|
+
For each asset confirmed available, decide where it belongs in the outline. Apply the spacing rule: no visual element adjacent to another. Images break up content within sections. CTAs go between sections, not within them.
|
|
42
|
+
|
|
43
|
+
### 5. Place Quotes Inline
|
|
44
|
+
|
|
45
|
+
Do not create a separate quotes section. Identify the 2–3 strongest quotes from the source material and mark where they belong inline within their relevant sections. Use blockquote format notation: `> [speaker name]: "quote text"`
|
|
46
|
+
|
|
47
|
+
### 6. Apply Recipe-Specific Conventions
|
|
48
|
+
|
|
49
|
+
If the recipe provides outline conventions for this content type, apply them now. Recipe overrides take precedence over general defaults when they conflict.
|
|
50
|
+
|
|
51
|
+
---
|
|
52
|
+
|
|
53
|
+
## Self-Review Checklist
|
|
54
|
+
|
|
55
|
+
Before producing output, verify:
|
|
56
|
+
|
|
57
|
+
1. Does each body section have exactly one core argument?
|
|
58
|
+
2. Are visual elements spaced — no two adjacent?
|
|
59
|
+
3. Are quotes marked inline, not in a separate section?
|
|
60
|
+
4. Is the word target reasonable given the number and depth of sections?
|
|
61
|
+
5. Does anything from the outline format's "What Does NOT Belong" section appear in the outline?
|
|
62
|
+
|
|
63
|
+
If any check fails, correct the outline before finalizing.
|
|
64
|
+
|
|
65
|
+
---
|
|
66
|
+
|
|
67
|
+
## Output Format
|
|
68
|
+
|
|
69
|
+
```
|
|
70
|
+
---
|
|
71
|
+
phase: outline
|
|
72
|
+
status: complete|needs_input
|
|
73
|
+
issue_key: {{from input}}
|
|
74
|
+
content_type: {{from recipe}}
|
|
75
|
+
word_count: {{actual word count of body below}}
|
|
76
|
+
next_phase: draft
|
|
77
|
+
needs_input_reason: {{only if status is needs_input}}
|
|
78
|
+
---
|
|
79
|
+
|
|
80
|
+
## Outline
|
|
81
|
+
|
|
82
|
+
**Title:** [proposed title]
|
|
83
|
+
|
|
84
|
+
**Intro**
|
|
85
|
+
- Hook: [1 sentence — the entry point that earns the reader's attention]
|
|
86
|
+
- Frame: [1–2 sentences — what this piece is about and why it matters now]
|
|
87
|
+
- Setup: [1 sentence — what the reader will get from reading]
|
|
88
|
+
|
|
89
|
+
**Body Sections** (3–5 sections)
|
|
90
|
+
|
|
91
|
+
### [Section Title]
|
|
92
|
+
- Core argument: [1 sentence]
|
|
93
|
+
- Key points: [bullet list]
|
|
94
|
+
- Assets: [asset placements, if any]
|
|
95
|
+
- Quotes: [inline quote placements, if any]
|
|
96
|
+
|
|
97
|
+
[repeat for each section]
|
|
98
|
+
|
|
99
|
+
**Related Reading**
|
|
100
|
+
- [3–5 links or resource types — carry from gather assets or note as TBD]
|
|
101
|
+
|
|
102
|
+
**Editorial Watch Items**
|
|
103
|
+
- [Things the draft phase should pay attention to: naming sensitivities, attribution requirements, gaps that need [bracket] treatment, recipe-specific requirements]
|
|
104
|
+
|
|
105
|
+
**Word Target:** [number from gather, adjusted if needed with reasoning]
|
|
106
|
+
```
|
|
@@ -0,0 +1,137 @@
|
|
|
1
|
+
# Phase 4: Review
|
|
2
|
+
|
|
3
|
+
You are the quality gate for a headless content pipeline. Your job is to score the draft, fix what you can fix without human input, and re-score to confirm improvement. What you cannot fix, you flag clearly for the editor.
|
|
4
|
+
|
|
5
|
+
This is a single-pass operation. Score, fix, re-score. Do not loop. Do not defer fixes to a later pass.
|
|
6
|
+
|
|
7
|
+
---
|
|
8
|
+
|
|
9
|
+
## Input Format
|
|
10
|
+
|
|
11
|
+
The consumer will provide:
|
|
12
|
+
|
|
13
|
+
- **Draft phase output** — complete draft with all structural elements
|
|
14
|
+
- **Outline phase output** — for fidelity checking
|
|
15
|
+
- **Recipe overrides** — content-type-specific review criteria (appended by consumer, if any)
|
|
16
|
+
- **Standards** — editorial standards injected by the consumer (voice, formatting, AI tells, etc.)
|
|
17
|
+
|
|
18
|
+
---
|
|
19
|
+
|
|
20
|
+
## Process
|
|
21
|
+
|
|
22
|
+
### 1. Initial Scoring
|
|
23
|
+
|
|
24
|
+
Score the draft across five categories. Use the grading format below. Be honest — inflated scores help nobody.
|
|
25
|
+
|
|
26
|
+
**Grading format:** Use exact emoji characters. Whole letter grades only (A, B, C, D, F — no + or − modifiers).
|
|
27
|
+
|
|
28
|
+
| Grade | Indicator |
|
|
29
|
+
|-------|-----------|
|
|
30
|
+
| A | 🟢 A |
|
|
31
|
+
| B | 🟢 B |
|
|
32
|
+
| C | 🟡 C |
|
|
33
|
+
| D | 🔴 D |
|
|
34
|
+
| F | 🔴 F |
|
|
35
|
+
|
|
36
|
+
**Categories:**
|
|
37
|
+
|
|
38
|
+
- **Brand Voice** — Sounds like IMA? Correct tone blend for format? No AI tells? No terminology violations?
|
|
39
|
+
- **Evidence Quality** — Claims backed by named studies or labeled as opinion? Precise language ("may help," not "cures")? No fabricated stats?
|
|
40
|
+
- **Audience Clarity** — Patient can understand it, physician won't cringe? Medical terms defined on first use? No unexplained jargon?
|
|
41
|
+
- **Structural Craft** — All required elements present and in order? Sections earn their place? Scannable? Outline fidelity maintained?
|
|
42
|
+
- **CTA Effectiveness** — CTAs specific and verb-first? Correctly formatted for channel? Reader knows what clicking will do?
|
|
43
|
+
|
|
44
|
+
### 2. Check Failure Modes
|
|
45
|
+
|
|
46
|
+
Identify any of these blocking issues before applying fixes:
|
|
47
|
+
|
|
48
|
+
- Missing required structural elements (compare against draft format standards)
|
|
49
|
+
- Outline fidelity failures — sections that deviated significantly from the approved outline
|
|
50
|
+
- Content-type compliance failures declared in the recipe
|
|
51
|
+
- Fabricated or unsupported health claims
|
|
52
|
+
- Terminology violations (e.g., "FLCCC" used, "cure" without qualification)
|
|
53
|
+
|
|
54
|
+
Document blocking issues separately — they require editor attention, not just a fix pass.
|
|
55
|
+
|
|
56
|
+
### 3. Fix What You Can
|
|
57
|
+
|
|
58
|
+
Apply fixes directly to the draft for:
|
|
59
|
+
|
|
60
|
+
- AI tells — replace with specific language from the editorial standards list
|
|
61
|
+
- Repeated phrases — vary or cut
|
|
62
|
+
- Missing bold phrases — add one bolded key phrase per H2 section
|
|
63
|
+
- Wall-of-text sections — add subheaders or break into shorter paragraphs
|
|
64
|
+
- Formatting issues — fix placeholder formatting, CTA format, header levels
|
|
65
|
+
- Weak paragraph openings — rewrite sentences that open with "There is," "It is," "This is"
|
|
66
|
+
- Vague references — make specific where source material supports it
|
|
67
|
+
- Minor voice drift — tune back to the Caregiver/Sage/Outlaw blend
|
|
68
|
+
|
|
69
|
+
Do not invent evidence. Do not restructure sections. Do not change the core argument of any section.
|
|
70
|
+
|
|
71
|
+
### 4. Re-Score
|
|
72
|
+
|
|
73
|
+
After applying fixes, re-score all five categories against the corrected draft. Document what changed and why.
|
|
74
|
+
|
|
75
|
+
---
|
|
76
|
+
|
|
77
|
+
## Self-Review Checklist
|
|
78
|
+
|
|
79
|
+
Before producing output, verify:
|
|
80
|
+
|
|
81
|
+
1. Is the initial scorecard honest — no inflated grades?
|
|
82
|
+
2. Were the fixes actually applied to the draft text, not just noted?
|
|
83
|
+
3. Is there a clear separation between "What I Fixed" and "What Needs Editor Attention"?
|
|
84
|
+
4. Does the re-score accurately reflect the corrected draft?
|
|
85
|
+
5. Are blocking issues documented with enough specificity for an editor to act on them?
|
|
86
|
+
|
|
87
|
+
If any check fails, correct before finalizing.
|
|
88
|
+
|
|
89
|
+
---
|
|
90
|
+
|
|
91
|
+
## Output Format
|
|
92
|
+
|
|
93
|
+
```
|
|
94
|
+
---
|
|
95
|
+
phase: review
|
|
96
|
+
status: complete|needs_input
|
|
97
|
+
issue_key: {{from input}}
|
|
98
|
+
content_type: {{from recipe}}
|
|
99
|
+
word_count: {{actual word count of corrected draft below}}
|
|
100
|
+
next_phase: deliver
|
|
101
|
+
needs_input_reason: {{only if status is needs_input}}
|
|
102
|
+
---
|
|
103
|
+
|
|
104
|
+
## Initial Scorecard
|
|
105
|
+
|
|
106
|
+
| Category | Grade | Notes |
|
|
107
|
+
|----------|-------|-------|
|
|
108
|
+
| Brand Voice | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
109
|
+
| Evidence Quality | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
110
|
+
| Audience Clarity | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
111
|
+
| Structural Craft | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
112
|
+
| CTA Effectiveness | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
113
|
+
|
|
114
|
+
## What I Fixed
|
|
115
|
+
|
|
116
|
+
- [specific fix applied]
|
|
117
|
+
- [specific fix applied]
|
|
118
|
+
|
|
119
|
+
## What Needs Editor Attention
|
|
120
|
+
|
|
121
|
+
- [blocking issue with enough specificity to act on]
|
|
122
|
+
- [blocking issue]
|
|
123
|
+
|
|
124
|
+
## Revised Scorecard
|
|
125
|
+
|
|
126
|
+
| Category | Grade | Notes |
|
|
127
|
+
|----------|-------|-------|
|
|
128
|
+
| Brand Voice | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
129
|
+
| Evidence Quality | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
130
|
+
| Audience Clarity | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
131
|
+
| Structural Craft | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
132
|
+
| CTA Effectiveness | 🟢/🟡/🔴 [letter] | [5–15 words] |
|
|
133
|
+
|
|
134
|
+
---
|
|
135
|
+
|
|
136
|
+
[Corrected draft — complete, with all fixes applied]
|
|
137
|
+
```
|
|
@@ -0,0 +1,159 @@
|
|
|
1
|
+
# Draft Format
|
|
2
|
+
|
|
3
|
+
Structural rules for drafts produced by the draft phase. Injected by the consumer when the recipe declares this file.
|
|
4
|
+
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
## What a Draft Is
|
|
8
|
+
|
|
9
|
+
A draft is a complete post blueprint that reads like the published page. Every structural element is in place. Every placeholder is marked with the exact placeholder format. An editor reading the draft should be able to move it to production with minimal changes.
|
|
10
|
+
|
|
11
|
+
A draft is not a partial document. It is not a summary of what will be written. It is not an outline with some prose filled in. If an element is missing, the draft is incomplete.
|
|
12
|
+
|
|
13
|
+
---
|
|
14
|
+
|
|
15
|
+
## 10 Required Elements
|
|
16
|
+
|
|
17
|
+
Produce all 10 elements in this order. Recipes may modify the sequence or add elements — follow recipe overrides when they conflict with this default.
|
|
18
|
+
|
|
19
|
+
### 1. Meta Description
|
|
20
|
+
|
|
21
|
+
150–160 characters. Keyword-rich. Describes what the reader will get from the page. Written as a complete sentence or two fragments. Does not begin with the site name.
|
|
22
|
+
|
|
23
|
+
```
|
|
24
|
+
Meta: [150–160 character description]
|
|
25
|
+
```
|
|
26
|
+
|
|
27
|
+
### 2. Title
|
|
28
|
+
|
|
29
|
+
Clear, specific, benefit-oriented. Not a topic label — a claim, a promise, or a question the reader is already asking. The title from the outline is a starting point; refine it for the draft.
|
|
30
|
+
|
|
31
|
+
```
|
|
32
|
+
# [Title]
|
|
33
|
+
```
|
|
34
|
+
|
|
35
|
+
### 3. Excerpt
|
|
36
|
+
|
|
37
|
+
The first paragraph of the piece, set apart as the excerpt. Must contain one bolded key phrase — the most important idea for a scanner to land on. The bolded phrase should be a noun phrase or short clause, not a single word.
|
|
38
|
+
|
|
39
|
+
```
|
|
40
|
+
**[bolded key phrase]** — [rest of excerpt paragraph]
|
|
41
|
+
```
|
|
42
|
+
|
|
43
|
+
Or the bold can fall mid-paragraph where it is most impactful.
|
|
44
|
+
|
|
45
|
+
### 4. Hero Element
|
|
46
|
+
|
|
47
|
+
One of:
|
|
48
|
+
- `[VIDEO EMBED: brief description of video content]`
|
|
49
|
+
- `[HERO IMAGE: brief description — subject, mood, source if known]`
|
|
50
|
+
|
|
51
|
+
Do not place a second visual element immediately after the hero. The prose intro follows.
|
|
52
|
+
|
|
53
|
+
### 5. Prose Intro
|
|
54
|
+
|
|
55
|
+
Two to three paragraphs that hook, frame, and set up the piece. Follows the hook/frame/setup structure from the outline's intro. Written in full prose — not bullets, not headers.
|
|
56
|
+
|
|
57
|
+
The intro should give the reader a reason to keep reading. It does not summarize the entire piece. It earns the reader's next scroll.
|
|
58
|
+
|
|
59
|
+
### 6. Citation Block
|
|
60
|
+
|
|
61
|
+
For content based on a study or published research, include the citation immediately after the intro:
|
|
62
|
+
|
|
63
|
+
```
|
|
64
|
+
**Citation:** [Author Last Names]. "[Study Title]." *[Journal Name]*. [Year]. [DOI or URL if available]
|
|
65
|
+
```
|
|
66
|
+
|
|
67
|
+
For content not based on a primary study, omit this element.
|
|
68
|
+
|
|
69
|
+
### 7. Body Sections
|
|
70
|
+
|
|
71
|
+
For each section from the outline, produce:
|
|
72
|
+
|
|
73
|
+
- An H2 header that states the section's core argument (not just the topic)
|
|
74
|
+
- Prose developing the argument, incorporating key points from the outline
|
|
75
|
+
- At least one bolded key phrase per section
|
|
76
|
+
- Inline blockquotes where the outline marked quote placements
|
|
77
|
+
- Asset placeholders where the outline marked asset placements
|
|
78
|
+
- A subheader (`###`) if the section covers more than two distinct sub-topics
|
|
79
|
+
|
|
80
|
+
No section should run more than four consecutive paragraphs without a visual break (image, quote, or subheader). Wall-of-text sections will be flagged in review.
|
|
81
|
+
|
|
82
|
+
Blockquote format:
|
|
83
|
+
```
|
|
84
|
+
> [Speaker Name]: "[near-verbatim quote]"
|
|
85
|
+
```
|
|
86
|
+
|
|
87
|
+
### 8. Related Reading
|
|
88
|
+
|
|
89
|
+
A list of three to five related resources with brief descriptions. Format:
|
|
90
|
+
|
|
91
|
+
```
|
|
92
|
+
## Related Reading
|
|
93
|
+
|
|
94
|
+
- **[Resource Title]** — [one sentence description] [URL or placeholder]
|
|
95
|
+
- **[Resource Title]** — [one sentence description] [URL or placeholder]
|
|
96
|
+
```
|
|
97
|
+
|
|
98
|
+
### 9. Forums Button
|
|
99
|
+
|
|
100
|
+
A call to action pointing to the IMA community forums. Use the standard placeholder:
|
|
101
|
+
|
|
102
|
+
```
|
|
103
|
+
[Forums Button]
|
|
104
|
+
```
|
|
105
|
+
|
|
106
|
+
Recipes may specify different community CTAs — follow the recipe override.
|
|
107
|
+
|
|
108
|
+
### 10. Asset Table
|
|
109
|
+
|
|
110
|
+
A reference block listing all embeds, downloads, and CTAs used in the piece. The editor uses this to confirm all assets are live before publishing.
|
|
111
|
+
|
|
112
|
+
```
|
|
113
|
+
## Asset Table
|
|
114
|
+
|
|
115
|
+
| Asset | Type | Location in Draft | Status |
|
|
116
|
+
|-------|------|-------------------|--------|
|
|
117
|
+
| [description] | Video embed | After intro | [URL or "TBD"] |
|
|
118
|
+
| [description] | Image | Section 2 | [URL or "TBD"] |
|
|
119
|
+
| [description] | PDF download | Related Reading | [URL or "TBD"] |
|
|
120
|
+
```
|
|
121
|
+
|
|
122
|
+
---
|
|
123
|
+
|
|
124
|
+
## Placeholder Conventions
|
|
125
|
+
|
|
126
|
+
Use these exact formats. Do not improvise placeholder syntax — the consumer may parse them.
|
|
127
|
+
|
|
128
|
+
| Placeholder | Use For |
|
|
129
|
+
|-------------|---------|
|
|
130
|
+
| `[VIDEO EMBED: description]` | Video that should be embedded |
|
|
131
|
+
| `[HERO IMAGE: description]` | Hero image at top of piece |
|
|
132
|
+
| `[STUDY SCREENSHOT: description]` | Figure or table from a study |
|
|
133
|
+
| `[Figure N: description]` | Any numbered figure |
|
|
134
|
+
| `[Interstitial: CTA name]` | Mid-content CTA (between sections) |
|
|
135
|
+
| `[Forums Button]` | Standard IMA forums CTA |
|
|
136
|
+
| `[specific missing data]` | Any specific fact not in source material — describe what is needed |
|
|
137
|
+
|
|
138
|
+
---
|
|
139
|
+
|
|
140
|
+
## Naming Rule
|
|
141
|
+
|
|
142
|
+
If the source material names it, the draft names it. There is no acceptable substitution of a specific reference for a generic one when the specific reference is available.
|
|
143
|
+
|
|
144
|
+
- Named study → use the study name
|
|
145
|
+
- Named drug or protocol → use the exact name as it appears in the source
|
|
146
|
+
- Named speaker → full name on first mention, last name only after
|
|
147
|
+
- Named institution → exact name, no abbreviation on first mention
|
|
148
|
+
|
|
149
|
+
If the source material does not name something, use the appropriate placeholder.
|
|
150
|
+
|
|
151
|
+
---
|
|
152
|
+
|
|
153
|
+
## What Does NOT Belong
|
|
154
|
+
|
|
155
|
+
- Body prose without structural elements — every piece of prose must live inside a named structural element
|
|
156
|
+
- Deferred decisions — "the editor can decide X" — make the decision and note it in editorial watch items if it is uncertain
|
|
157
|
+
- Generic references when specific ones are available in the source material
|
|
158
|
+
- Assets that have not been seen or confirmed — use a placeholder, not an invented description
|
|
159
|
+
- Prose summaries of what sections will contain — write the sections, do not summarize them
|
package/plugins/ima-claude/skills/agentic-workflows/references/standards/editorial-standards.md
ADDED
|
@@ -0,0 +1,160 @@
|
|
|
1
|
+
# Editorial Standards
|
|
2
|
+
|
|
3
|
+
Universal quality criteria for IMA content. Injected by the consumer when the recipe declares this file.
|
|
4
|
+
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
## Voice Archetypes
|
|
8
|
+
|
|
9
|
+
IMA content blends three archetypes simultaneously. Do not use them in turns or in separate sections — the blend must be present in every paragraph.
|
|
10
|
+
|
|
11
|
+
| Archetype | What It Brings | How It Shows |
|
|
12
|
+
|-----------|---------------|--------------|
|
|
13
|
+
| **Caregiver** | Warmth, solidarity, human stakes | Acknowledges what patients are living with; writes to an ally, not a target |
|
|
14
|
+
| **Sage** | Evidence, precision, credibility | Names studies, cites journals, qualifies claims accurately |
|
|
15
|
+
| **Outlaw** | Institutional critique, moral clarity | Names specific actors when warranted; earned intensity, not reflexive antagonism |
|
|
16
|
+
|
|
17
|
+
The blend should be invisible. Readers should feel it, not label it.
|
|
18
|
+
|
|
19
|
+
---
|
|
20
|
+
|
|
21
|
+
## Writing Principles
|
|
22
|
+
|
|
23
|
+
### Lead with Progress
|
|
24
|
+
|
|
25
|
+
Open with what has been learned, discovered, or achieved — not with the problem alone. The reader should leave the opening paragraph feeling informed, not alarmed.
|
|
26
|
+
|
|
27
|
+
### Organize Around Principles
|
|
28
|
+
|
|
29
|
+
Structure body sections around the insights a reader can act on or carry forward. Do not organize around the source material's sequence unless that sequence is the best reader sequence.
|
|
30
|
+
|
|
31
|
+
### Quotes Near-Verbatim
|
|
32
|
+
|
|
33
|
+
Quote speakers with fidelity to their actual words. Minor grammatical cleanup is acceptable. Do not paraphrase a direct quote and present it as a quote. If you cannot quote exactly, use indirect attribution.
|
|
34
|
+
|
|
35
|
+
### Never Fabricate
|
|
36
|
+
|
|
37
|
+
Do not invent statistics, study names, speaker quotes, dates, or institutional details. If information is not in the source material, use a placeholder: `[specific data needed]`. Placeholders are preferable to fabrications.
|
|
38
|
+
|
|
39
|
+
### Evergreen Where Possible
|
|
40
|
+
|
|
41
|
+
Write in a way that will not be immediately dated. Avoid "this week," "recently," "just published" — use the specific date or publication name instead.
|
|
42
|
+
|
|
43
|
+
### Speaker Attribution
|
|
44
|
+
|
|
45
|
+
Name speakers on first mention with full name and title. Use last name only on subsequent mentions within the same section. Do not refer to speakers as "the expert" or "the researcher" — use their name.
|
|
46
|
+
|
|
47
|
+
### Conviction, Not Dramatization
|
|
48
|
+
|
|
49
|
+
State positions directly when the evidence supports them. Do not manufacture urgency through exclamation points, all-caps, or breathless phrasing. Earned intensity comes from the weight of the evidence, not from theatrical framing.
|
|
50
|
+
|
|
51
|
+
### Strongest Findings Proportional
|
|
52
|
+
|
|
53
|
+
Lead with the findings that are most supported and most significant. Do not bury the main finding in a middle section. If the study's central finding is X, X should appear in the intro or the first body section.
|
|
54
|
+
|
|
55
|
+
### Cumulative Burden Framing
|
|
56
|
+
|
|
57
|
+
When presenting multiple findings, build toward a conclusion. Do not list findings as isolated data points. Connect them: this plus this plus this means something.
|
|
58
|
+
|
|
59
|
+
---
|
|
60
|
+
|
|
61
|
+
## Formatting Rules
|
|
62
|
+
|
|
63
|
+
### Markdown
|
|
64
|
+
|
|
65
|
+
Use pure markdown. No HTML. No inline styles.
|
|
66
|
+
|
|
67
|
+
### Colons in Lists
|
|
68
|
+
|
|
69
|
+
Use colons to introduce lists: `**Key finding:** explanation`. Do not use en-dashes or em-dashes as list introducers.
|
|
70
|
+
|
|
71
|
+
### Em-Dashes
|
|
72
|
+
|
|
73
|
+
Use sparingly — only when a comma or parenthetical won't do. Not more than two per piece.
|
|
74
|
+
|
|
75
|
+
### One Idea Per Visual Block
|
|
76
|
+
|
|
77
|
+
Each image, callout, or embedded asset should stand alone visually. Do not stack two visual elements without prose between them.
|
|
78
|
+
|
|
79
|
+
### Bold One Phrase Per Section
|
|
80
|
+
|
|
81
|
+
Every H2 section should have exactly one bolded key phrase — the most important takeaway for a scanner. Do not bold entire sentences or multiple phrases per section.
|
|
82
|
+
|
|
83
|
+
### Bold Keyword in H2s
|
|
84
|
+
|
|
85
|
+
H2 section headers should contain the most important keyword for that section. Write headers that communicate the section's core argument, not just its topic.
|
|
86
|
+
|
|
87
|
+
### No Unearned Jargon
|
|
88
|
+
|
|
89
|
+
Define any medical term, acronym, or technical concept on first use. Format: `[term] ([brief plain-language definition])`. Do not assume the reader knows clinical terminology.
|
|
90
|
+
|
|
91
|
+
### Single-Sentence Paragraphs
|
|
92
|
+
|
|
93
|
+
Acceptable when the sentence carries enough weight to stand alone. Not acceptable as padding or as a substitute for developing an argument.
|
|
94
|
+
|
|
95
|
+
### 3+ Findings = Bullet List
|
|
96
|
+
|
|
97
|
+
When presenting three or more discrete findings, use a bullet list. Prose strings of findings are harder to scan and easier to misread.
|
|
98
|
+
|
|
99
|
+
### Multiple Sub-Topics = Bold Sub-Headers + Preview List
|
|
100
|
+
|
|
101
|
+
When a section covers more than two distinct sub-topics, introduce them with a brief preview list before the sub-headers. This helps scanners navigate.
|
|
102
|
+
|
|
103
|
+
---
|
|
104
|
+
|
|
105
|
+
## AI Tells to Avoid
|
|
106
|
+
|
|
107
|
+
These phrases signal AI-generated content and must not appear in any draft:
|
|
108
|
+
|
|
109
|
+
- "In today's rapidly evolving..."
|
|
110
|
+
- "In the realm of..."
|
|
111
|
+
- "It's worth noting that..."
|
|
112
|
+
- "It is important to note that..."
|
|
113
|
+
- "At its core,..."
|
|
114
|
+
- "Let's delve into..."
|
|
115
|
+
- "This begs the question..."
|
|
116
|
+
- "In conclusion,..."
|
|
117
|
+
- "Moreover,..." (as a paragraph opener — occasional mid-sentence use is fine)
|
|
118
|
+
- "Furthermore,..." (as a paragraph opener)
|
|
119
|
+
- "That being said,..."
|
|
120
|
+
- "Navigate the complexities of..."
|
|
121
|
+
- "Unlock the potential of..."
|
|
122
|
+
- "Leverage" (when "use" will do)
|
|
123
|
+
- "Robust" (as a generic intensifier)
|
|
124
|
+
- "Comprehensive" (as a generic intensifier)
|
|
125
|
+
- "Innovative" (without specific description of what is new)
|
|
126
|
+
- "Cutting-edge" (without specific description)
|
|
127
|
+
- "Seamless" (as a generic positive)
|
|
128
|
+
- "Deep dive"
|
|
129
|
+
- "Unpack"
|
|
130
|
+
- "Explore" (as a verb for what the piece will do — use "examine," "analyze," or "show")
|
|
131
|
+
|
|
132
|
+
If three or more AI tells appear in a draft, the Brand Voice score drops one letter grade.
|
|
133
|
+
|
|
134
|
+
---
|
|
135
|
+
|
|
136
|
+
## IMA Terminology
|
|
137
|
+
|
|
138
|
+
| Avoid | Use Instead | Why |
|
|
139
|
+
|-------|-------------|-----|
|
|
140
|
+
| FLCCC | IMA / Independent Medical Alliance | Rebrand complete |
|
|
141
|
+
| "Cure" (unqualified) | "Protocol," "approach," "may help with" | Precision |
|
|
142
|
+
| "Treatment" (unqualified) | "Protocol," "approach" | Precision |
|
|
143
|
+
| "You should" | "You may consider," "Research suggests" | Patient autonomy |
|
|
144
|
+
| "Alternative medicine" | "Integrative medicine," "complementary approaches" | Accurate framing |
|
|
145
|
+
| "Big Pharma" (as sole descriptor) | Name the company, or "pharmaceutical industry" | Credibility |
|
|
146
|
+
| "We believe" (for science) | "Evidence shows," "Research indicates" | Evidence-based positioning |
|
|
147
|
+
| "Anti-vaccine" (describing IMA) | "Pro-informed-consent," "independent" | Accuracy |
|
|
148
|
+
|
|
149
|
+
---
|
|
150
|
+
|
|
151
|
+
## Independence Signal
|
|
152
|
+
|
|
153
|
+
At least one of the following must appear in any long-form piece:
|
|
154
|
+
|
|
155
|
+
- "Independent" in the organizational description
|
|
156
|
+
- "No pharma funding" or equivalent
|
|
157
|
+
- "501(c)(3) nonprofit" identification
|
|
158
|
+
- "Funded by people" / "funded by donors" language
|
|
159
|
+
|
|
160
|
+
Social posts are exempt — the profile bio carries this. Op-eds and statements are not exempt.
|
|
@@ -0,0 +1,110 @@
|
|
|
1
|
+
# Outline Format
|
|
2
|
+
|
|
3
|
+
Structural rules for outlines produced by the outline phase. Injected by the consumer when the recipe declares this file.
|
|
4
|
+
|
|
5
|
+
---
|
|
6
|
+
|
|
7
|
+
## What an Outline Is
|
|
8
|
+
|
|
9
|
+
An outline is a structural proposal in bullet points. It is not a draft. It contains no prose paragraphs. It communicates architecture — what sections exist, in what order, what each section argues, and where visual elements land.
|
|
10
|
+
|
|
11
|
+
The outline is the contract between the outline phase and the draft phase. The draft phase follows the outline unless a deviation is explicitly documented.
|
|
12
|
+
|
|
13
|
+
---
|
|
14
|
+
|
|
15
|
+
## Governing Principle
|
|
16
|
+
|
|
17
|
+
Structure around reader needs, not source material architecture.
|
|
18
|
+
|
|
19
|
+
The source material (transcript, study, press release) has its own sequence driven by how it was produced. That sequence is rarely the best sequence for a reader encountering the material for the first time. The outline phase's job is to find the reader's sequence.
|
|
20
|
+
|
|
21
|
+
Ask: what does the reader need to know first to understand what comes next? Start there.
|
|
22
|
+
|
|
23
|
+
---
|
|
24
|
+
|
|
25
|
+
## Required Elements
|
|
26
|
+
|
|
27
|
+
Every outline must contain these elements in this order:
|
|
28
|
+
|
|
29
|
+
### Title
|
|
30
|
+
A proposed headline. Clear, specific, benefit-oriented. Not a topic label — a claim or a promise. The draft phase may refine it, but the outline provides a working title.
|
|
31
|
+
|
|
32
|
+
### Intro
|
|
33
|
+
Three components:
|
|
34
|
+
- **Hook** — the entry point that earns the reader's attention. One sentence. Can be a bold claim, a striking finding, a tension, or a question the reader is already asking.
|
|
35
|
+
- **Frame** — what this piece is about and why it matters now. One to two sentences.
|
|
36
|
+
- **Setup** — what the reader will get from reading. One sentence.
|
|
37
|
+
|
|
38
|
+
### Body Sections (3–5)
|
|
39
|
+
For each section:
|
|
40
|
+
- **Section title** — communicates the core argument, not just the topic
|
|
41
|
+
- **Core argument** — one sentence stating what this section claims
|
|
42
|
+
- **Key points** — bullet list of the specific evidence, data, or ideas that support the core argument
|
|
43
|
+
- **Asset placements** — where images, video embeds, or CTAs land within this section (if applicable)
|
|
44
|
+
- **Quote placements** — where blockquotes land inline, with the speaker and a fragment of the quote
|
|
45
|
+
|
|
46
|
+
Sections must not overlap in argument. If two sections are making the same point, merge them.
|
|
47
|
+
|
|
48
|
+
### Related Reading
|
|
49
|
+
Three to five links or resource types. Can be TBD if assets are not yet confirmed. Include a brief description of each.
|
|
50
|
+
|
|
51
|
+
### Editorial Watch Items
|
|
52
|
+
Things the draft phase should pay special attention to:
|
|
53
|
+
- Naming sensitivities (speakers, institutions, studies)
|
|
54
|
+
- Attribution requirements (quotes that must be near-verbatim)
|
|
55
|
+
- Gaps that need `[bracket]` treatment
|
|
56
|
+
- Recipe-specific requirements the draft phase must apply
|
|
57
|
+
- Anything from the gather phase gaps section that was not resolved
|
|
58
|
+
|
|
59
|
+
### Word Target
|
|
60
|
+
Carry forward from the gather phase. If the outline reveals the scope is larger or smaller than estimated, adjust and explain the change in one sentence.
|
|
61
|
+
|
|
62
|
+
---
|
|
63
|
+
|
|
64
|
+
## Section Scope
|
|
65
|
+
|
|
66
|
+
Each body section argues exactly one thing. The section title states that argument. Every key point in the section supports that argument. If a section has key points that pull in two directions, split it into two sections.
|
|
67
|
+
|
|
68
|
+
Aim for 3–5 sections. Fewer than 3 means the content lacks depth. More than 5 means the scope is too broad or the arguments are not being consolidated.
|
|
69
|
+
|
|
70
|
+
---
|
|
71
|
+
|
|
72
|
+
## Visual Element Placement
|
|
73
|
+
|
|
74
|
+
Apply these rules for every asset:
|
|
75
|
+
|
|
76
|
+
- **No adjacent visual elements** — never place two images, embeds, or CTAs in sequence without prose between them
|
|
77
|
+
- **Images break up content within sections** — place images mid-section when a section runs long, not at the top as a header image (unless the recipe specifies a hero)
|
|
78
|
+
- **CTAs go between sections** — interstitial CTAs belong in the transition between sections, not embedded within a section's prose
|
|
79
|
+
- **Video embeds** — typically at the top of the piece as the hero element; note this in the intro element, not a body section
|
|
80
|
+
|
|
81
|
+
---
|
|
82
|
+
|
|
83
|
+
## Quote Placement
|
|
84
|
+
|
|
85
|
+
Quotes belong inline where they would naturally land if the reader encountered them in prose — not in a separate "Quotes" section.
|
|
86
|
+
|
|
87
|
+
For each quote, mark its placement in the relevant section's key points:
|
|
88
|
+
|
|
89
|
+
```
|
|
90
|
+
- Quote: [Speaker last name]: "[fragment of quote]" — after point about [topic]
|
|
91
|
+
```
|
|
92
|
+
|
|
93
|
+
Blockquote format in the draft:
|
|
94
|
+
```
|
|
95
|
+
> [Speaker Name]: "exact or near-verbatim quote"
|
|
96
|
+
```
|
|
97
|
+
|
|
98
|
+
Select quotes that advance the section's core argument. Do not include quotes for variety or because they seem interesting. Every quote should be the best available evidence for the point it follows.
|
|
99
|
+
|
|
100
|
+
---
|
|
101
|
+
|
|
102
|
+
## What Does NOT Belong in an Outline
|
|
103
|
+
|
|
104
|
+
- Prose paragraphs — outlines use bullets, not paragraphs
|
|
105
|
+
- Every detail from the source material — only what advances the reader's understanding
|
|
106
|
+
- Meta descriptions — these belong in the draft
|
|
107
|
+
- A separate "Quotes" section — quotes go inline
|
|
108
|
+
- The source material's own section structure — unless it happens to be the best reader sequence
|
|
109
|
+
- Stacked visual elements — no two adjacent assets
|
|
110
|
+
- Speculative sections — if you are not confident a section belongs, it does not belong
|