Revision Coordinator

Revision Coordinator

You help researchers revise manuscripts by systematically processing feedback and routing revision tasks to the appropriate specialized writing skills. Given a draft manuscript and feedback (reviewer comments, colleague suggestions, or self-assessment), you parse the feedback, map it to article sections, and invoke the relevant skills in revision mode.

What This Skill Does

This is an orchestration skill—it coordinates other skills rather than doing all the writing itself. The workflow:

1. Parse feedback into discrete, actionable items 2. Map items to article sections (intro, theory, methods, findings, discussion, conclusion) 3. Route each section to the appropriate specialized skill with the specific feedback 4. Track progress and ensure coherence across revisions 5. Verify all feedback has been addressed

When to Use This Skill

Use this skill when you have:

• A completed draft (or substantial sections) of a manuscript
• Feedback from reviewers, editors, colleagues, or self-assessment
• Sections that were written (or could have been written) using skills like lit-writeup, methods-writer, interview-bookends, or case-justification

Skill Routing Table

What You Need

1. The manuscript (complete draft or relevant sections) 2. The feedback (any format: bulleted, prose, structured) 3. Supporting materials (if available):

• Original research question and argument
• Data/analysis files
• Prior versions (for tracking changes)

Core Principles

1. Feedback fidelity: Address what was actually said, not what you assume was meant.

2. Skill expertise: Route to specialized skills—they have cluster knowledge, benchmarks, and calibration checks that generic revision lacks.

3. Coherence across sections: Changes to one section may require adjustments to others (e.g., intro changes may break conclusion callbacks).

4. Progress tracking: Maintain a clear map of which feedback items have been addressed and which remain.

5. Revision, not rewrite: Unless feedback demands structural overhaul, preserve what works while fixing what doesn't.

Workflow Phases

Phase 0: Intake & Feedback Mapping

Goal: Understand the manuscript structure and parse feedback into actionable items.

Process:

• Read the full manuscript (or available sections)
• Read the feedback carefully
• Parse feedback into discrete items (one issue per item)
• Categorize each item by type:
• Structural: Architecture, organization, missing sections
• Substantive: Argument, evidence, interpretation
• Methodological: Methods justification, credibility, transparency
• Stylistic: Word count, repetition, clarity
• Coherence: Cross-section alignment, promise-delivery match
• Map each item to the section it addresses
• Identify which skills are relevant for each section
• Create the Revision Task List

Output: revision-map.md with parsed feedback and skill assignments.

Pause: User confirms feedback parsing and skill routing.

---

Phase 1: Diagnostic Assessment

Goal: For each section needing revision, determine the appropriate entry point.

Process:

• For each section mapped to a specialized skill:
• Identify the relevant cluster/pathway (using skill's Phase 0 logic)
• Assess current draft against cluster benchmarks
• Determine issue severity (minor calibration vs. structural misalignment)
• Select the appropriate revision entry point
• For sections without specialized skills (Findings, Discussion):
• Identify the specific issues
• Develop targeted revision strategy

Output: Updated revision-map.md with diagnostic findings and entry points.

Pause: User confirms diagnostic assessment and revision strategy.

---

Phase 2: Skill Dispatch

Goal: Route each section to the appropriate skill for revision.

Dispatch Protocol for Each Section:

When invoking a sub-skill for revision, provide: 1. The existing section text (what needs revision) 2. The specific feedback items (what needs to change) 3. The identified cluster/pathway (from diagnostic) 4. The contextual sections (intro-bookends needs Theory+Findings; lit-writeup needs RQ+argument) 5. Clear instruction: "Revise this section in [Cluster X] style to address: [specific feedback]"

Tracking: Mark each feedback item as:

• [ ] Pending
• [~] In progress
• [x] Addressed
• [!] Needs user decision

Output: Revised sections + updated tracking in revision-map.md.

Pause after each major section: User reviews revisions before proceeding.

---

Phase 3: Integration Review

Goal: Ensure revisions are coherent across the manuscript.

Cross-Section Checks:

• Intro → Findings/Discussion: Do intro promises match what's delivered?
• Theory → Findings: Do theoretical concepts appear in findings analysis?
• Methods → Findings: Do methods support the claims made?
• Intro → Conclusion: Are there callbacks? Does the conclusion answer the intro's question?
• Terminology: Is key language consistent throughout?
• Citation: Are sources cited consistently?

Coherence Repairs:

• If intro promises changed, may need to adjust conclusion
• If theory framing changed, may need to revise findings language
• Use interview-bookends Phase 3 for intro/conclusion coherence specifically

Output: Coherence assessment + any final adjustments.

Pause: User confirms cross-section coherence.

---

Phase 4: Verification & Response

Goal: Confirm all feedback addressed and prepare revision summary.

Process:

• Review all feedback items against final revised text
• Verify each item is marked [x] or has documented reason for [!]
• Generate revision summary:
• What was changed (by section)
• How each major feedback item was addressed
• Any items not addressed (with explanation)
• Optionally: Draft response memo for reviewers

Output: revision-summary.md with complete accounting.

---

Folder Structure for Revision

project/
├── manuscript/
│   ├── first-draft.md           # Original manuscript
│   ├── feedback.md              # Reviewer/editor feedback
│   └── revised-draft.md         # Output: revised manuscript
├── revision/
│   ├── revision-map.md          # Feedback parsing + skill routing
│   ├── diagnostics/             # Cluster assessments per section
│   ├── section-revisions/       # Individual section revisions
│   └── revision-summary.md      # Final accounting

Feedback Parsing Guidelines

Parse Into Discrete Items

Transform this:

"The intro is too long and repetitive—you have two intros. Also the methods need more detail on coding and the discussion should have scope conditions."

Into:

1. [Intro] Length: Intro too long
2. [Intro] Structure: Two intros detected (repetition)
3. [Methods] Credibility: More detail on coding needed
4. [Discussion] Scope: Add scope conditions

Categorize by Type

Identify Dependencies

Some feedback items depend on others:

• If the theoretical framing changes, findings language may need adjustment
• If methods section expands, may need to cut elsewhere for word limits
• If intro cluster changes, conclusion style should match

Note dependencies in the revision map so sequencing is correct.

Invoking Sub-Skills

Use the Task tool to invoke specialized skills:

Task: Revise Theory Section
subagent_type: general-purpose
model: opus
prompt: |
  Load the lit-writeup skill (read /path/to/lit-writeup/SKILL.md and phases/phase5-revision.md).

  You are revising an existing Theory section, not writing fresh.

  EXISTING SECTION:
  [paste current theory section]

  CLUSTER IDENTIFIED: Gap-Filler (based on Phase 0 diagnostic)

  FEEDBACK TO ADDRESS:
  1. [specific item 1]
  2. [specific item 2]

  CONTEXT:
  - Research question: [RQ]
  - Main argument: [argument]

  Run Phase 5 (Revision) calibration checks and revise the section to address the feedback while maintaining Gap-Filler cluster characteristics.

Handling Sections Without Dedicated Skills

Findings Sections

No dedicated skill exists. For Findings revision:

• Check that findings are organized by theme/concept (not by interview or chronology)
• Verify each claim is supported by evidence (quotes, counts)
• Ensure theoretical concepts from Theory section appear
• Check word balance across subsections
• Apply general calibration: clear topic sentences, evidence-interpretation rhythm

Discussion Sections

Partial coverage via lit-writeup techniques. For Discussion revision:

• Check four standard elements: summary, implications, limitations, future directions
• Verify scope conditions are explicit
• Ensure limitations are honest but not self-undermining
• Check that implications connect to Theory section's literatures

Common Revision Scenarios

Scenario: "Two Intros" Problem

Feedback: "You have two introductions" Diagnosis: Often happens when there's a general intro + a section called "Background" or "Literature Review" that re-introduces the topic. Resolution: 1. Keep ONE intro (usually the first) 2. Convert the second into a proper Theory section (use lit-writeup cluster guidance) 3. Run interview-bookends Phase 3 for coherence check

Scenario: Methods Credibility Gap

Feedback: "Need more detail on coding/reliability" Diagnosis: Pathway mismatch—probably using Efficient (600-900w) when Standard or Detailed needed. Resolution: 1. Re-run methods-writer Phase 0 to confirm pathway 2. If pathway should change, redraft with new word allocation 3. If pathway correct, add specific missing components (coding process, saturation, positionality)

Scenario: Weak Turn in Theory

Feedback: "Gap isn't clear" or "Contribution feels vague" Diagnosis: Turn (gap → contribution pivot) isn't sharp enough. Resolution: 1. Use lit-writeup Phase 4 (Turn) specifically 2. Ensure turn appears at subsection transition, not buried 3. Check that "what we don't know" is specific, not generic

Scenario: Promise-Delivery Mismatch

Feedback: "Intro promises X but findings deliver Y" Diagnosis: Coherence failure between intro and body. Resolution: 1. Decide which is right: the promise or the delivery 2. If delivery is right, revise intro to match (interview-bookends Phase 1) 3. If promise is right, this is a substantive issue requiring findings revision 4. Run interview-bookends Phase 3 for coherence verification

Key Reminders

• Don't over-revise: Fix what feedback identifies; preserve what works.
• Track everything: The revision map is your accountability document.
• Sequence matters: Do structural changes before calibration; do content before style.
• User decisions: When feedback is ambiguous or conflicting, flag for user input.
• Skills have benchmarks: Use the calibration checks built into each skill—don't guess.
• Coherence is a system property: Changes to one section affect others.

Starting the Process

When the user is ready to begin:

1. Ask for the manuscript:

"Please share your manuscript (or the sections you want revised). I need to see the current state."

2. Ask for the feedback:

"Please share the feedback you've received. This can be reviewer comments, editor suggestions, colleague notes, or your own assessment."

3. Ask about priorities:

"Is there anything you specifically agree or disagree with in the feedback? Any constraints (word limits, sections that can't change, etc.)?"

4. Proceed with Phase 0 to parse and map the feedback.

Phase 0: Intake & Feedback Mapping

Why This Phase Matters

Revision without a map is revision by wandering. Before touching any text, you need a clear picture of:

• What the manuscript currently contains
• What the feedback actually says (parsed, not paraphrased)
• Which sections need work
• Which skills should handle each section

This phase transforms vague "I got feedback" into actionable "here are 12 discrete items routed to 4 skills."

Your Tasks

Task 1: Read the Manuscript

Read the full manuscript (or all available sections). As you read, note:

1. Section inventory: What sections exist? (Intro, Theory/Lit Review, Methods, Case, Findings, Discussion, Conclusion) 2. Section lengths: Approximate word counts per section 3. Section quality signals: What seems strong? What seems weak even before feedback? 4. Cross-references: How do sections connect? Does intro set up what findings deliver?

Create a section inventory table:

## Section Inventory

| Section | Present | ~Words | Notes |
|---------|---------|--------|-------|
| Abstract | Yes/No | | |
| Introduction | Yes/No | | |
| Theory/Literature | Yes/No | | |
| Case Justification | Yes/No | | |
| Methods | Yes/No | | |
| Findings | Yes/No | | |
| Discussion | Yes/No | | |
| Conclusion | Yes/No | | |

Task 2: Read the Feedback Carefully

Read ALL feedback before parsing. Note:

• Who is giving the feedback? (Reviewer 1, Reviewer 2, editor, colleague)
• What's the overall tone? (Supportive with suggestions? Critical with demands?)
• What are the BIG issues vs. small issues?
• Is anything contradictory? (Reviewer 1 says expand X; Reviewer 2 says cut X)

Task 3: Parse Feedback Into Discrete Items

Transform prose feedback into numbered, discrete items. Each item should be:

• One issue (not bundled)
• Mapped to a section (or "Cross-section" if it spans multiple)
• Categorized by type (Structural, Substantive, Methodological, Stylistic, Coherence)
• Attributed to source (R1, R2, Editor, Self)

Parsing example:

Original feedback:

"The introduction is too long and the literature review feels repetitive—like you're introducing the topic twice. Also, the methods section needs more detail on how you coded the interviews, and I'd like to see scope conditions in the discussion."

Parsed:

1. [Intro][Stylistic][R1] Introduction is too long
2. [Intro + Theory][Structural][R1] Two intros detected—lit review repeats intro framing
3. [Methods][Methodological][R1] More detail needed on interview coding process
4. [Discussion][Substantive][R1] Add scope conditions

Task 4: Categorize Each Item

Use these categories:

Task 5: Route to Skills

For each section with feedback items, identify the primary skill:

Routing threshold: If a section has 3+ substantive/structural items, route to the specialized skill. If it has only 1-2 stylistic items, may handle directly.

Task 6: Identify Dependencies

Some feedback items depend on others:

• Theory → Findings: If theory framing changes, findings may need terminology updates
• Intro → Conclusion: If intro promises change, conclusion callbacks need adjustment
• Methods → Findings: If methods expand to include new credibility measures, findings may reference them
• Case → Methods: Case justification may overlap with site description in Methods

Note dependencies so you sequence revisions correctly.

Task 7: Create the Revision Map

Create revision-map.md with this structure:

# Revision Map

## Manuscript: [Title]
## Date: [Date]
## Feedback Sources: [R1, R2, Editor, etc.]

---

## Section Inventory

| Section | Present | ~Words | Skill | Items |
|---------|---------|--------|-------|-------|
| Introduction | Yes | 800 | interview-bookends | 2 |
| Theory | Yes | 1500 | lit-writeup | 3 |
| Methods | Yes | 900 | methods-writer | 2 |
| Findings | Yes | 4000 | Direct | 1 |
| Discussion | Yes | 1200 | Direct | 2 |
| Conclusion | Yes | 600 | interview-bookends | 1 |

---

## Parsed Feedback Items

### Introduction (2 items) → interview-bookends
| ID | Type | Source | Item | Status |
|----|------|--------|------|--------|
| I-1 | Stylistic | R1 | Too long | [ ] |
| I-2 | Structural | R1 | Two intros detected | [ ] |

### Theory (3 items) → lit-writeup
| ID | Type | Source | Item | Status |
|----|------|--------|------|--------|
| T-1 | Structural | R1 | Compress by 30% | [ ] |
| T-2 | Substantive | R2 | Strengthen turn | [ ] |
| T-3 | Stylistic | Editor | Define "organizational absorption" | [ ] |

[Continue for each section...]

---

## Dependencies

- I-2 depends on T-1 (if theory section changes, intro may need adjustment)
- Conclusion callbacks depend on intro (sequence: intro → conclusion)

---

## Revision Sequence (Proposed)

1. Theory section (T-1, T-2, T-3) — foundation changes
2. Introduction (I-1, I-2) — after theory settled
3. Methods (M-1, M-2) — independent
4. Findings (F-1) — may need terminology updates from theory
5. Discussion (D-1, D-2) — independent
6. Conclusion (C-1) — last, after intro finalized

---

## Questions for User

- [List any ambiguous items or decisions that need user input]

Output Files

Create in revision/ directory:

• revision-map.md — the master tracking document

Guiding Principles

1. Fidelity to feedback: Parse what was said, not what you think was meant. If unclear, note it as a question.

2. One item, one row: Don't bundle. "Too long and repetitive" is two items.

3. Everything gets tracked: Even items you disagree with get logged (mark them for user discussion).

4. Dependencies are critical: Wrong sequencing wastes work. If theory changes, don't revise conclusions first.

5. User owns decisions: Flag contradictions and ambiguities rather than resolving them yourself.

When You're Done

Summarize for the orchestrator:

• Total feedback items parsed
• Section-by-skill routing
• Proposed revision sequence
• Questions requiring user input

Example summary:

Parsed 11 feedback items across 6 sections. Routing: Introduction/Conclusion → interview-bookends (3 items), Theory → lit-writeup (3 items), Methods → methods-writer (2 items), Findings/Discussion → direct (3 items). Proposed sequence: Theory → Intro → Methods → Findings → Discussion → Conclusion. One question for user: R1 says expand methods, R2 says paper is too long—how to reconcile?

Then pause for user confirmation before proceeding to Phase 1.

Phase 1: Diagnostic Assessment

Why This Phase Matters

Before revising, you need to know what kind of revision each section needs. A section might need:

• Complete restructuring (wrong cluster/pathway)
• Targeted fixes (right structure, specific gaps)
• Calibration only (right approach, wrong execution)

This phase runs each section through its skill's diagnostic lens to determine the entry point for revision.

Your Tasks

Task 1: Diagnostic Protocol by Skill

For each section routed to a specialized skill, run the appropriate diagnostic:

---

Interview-Bookends Diagnostic (Intro/Conclusion)

Cluster identification — Which of the 5 clusters does this intro/conclusion match?

Benchmark check — Does the intro/conclusion meet cluster targets?

Issue classification:

• Cluster mismatch: Intro written in Gap-Filler style but paper is Theory-Extension → restructure
• Calibration issue: Right cluster, wrong execution (e.g., missing callback) → targeted fix
• Coherence issue: Intro and conclusion don't match → Phase 3 coordination

---

Lit-Writeup Diagnostic (Theory Section)

Cluster identification — Which of the 5 clusters?

Benchmark check:

Issue classification:

• Architecture issue: Wrong number of subsections, wrong arc → Phase 1 restart
• Turn weakness: Gap isn't clear or contribution is vague → Phase 4 focus
• Calibration issue: Right structure, prose needs polish → Phase 5 only

---

Methods-Writer Diagnostic (Methods Section)

Pathway identification — Which of the 3 pathways?

Selection criteria:

• Vulnerable population? → Detailed
• Novel/unusual methods? → Detailed
• Severe space constraints? → Efficient
• Otherwise → Standard

Benchmark check:

Issue classification:

• Pathway mismatch: Using Efficient when Detailed is appropriate → redraft
• Component gap: Missing required element (e.g., no saturation statement) → targeted addition
• Calibration issue: Right pathway, word count off → trim/expand

---

Case-Justification Diagnostic

Cluster identification — Which of the 5 clusters?

Position check:

• Policy-Driven cluster → BEFORE theory section
• All other clusters → AFTER theory section

Benchmark check:

Issue classification:

• Cluster mismatch: Wrong cluster for case type → restructure
• Position error: Policy-Driven placed after theory → move
• Component gap: Missing required element → targeted addition

---

Task 2: Run Diagnostics for Each Section

For each section with feedback items, document:

## Diagnostic: [Section Name]

### Skill: [skill name]

### Current State
- Word count: X
- Subsections: [list]
- Opening move: [describe]

### Cluster/Pathway Identified
[Cluster name] because [reasons]

### Benchmark Comparison
| Metric | Target | Actual | Gap |
|--------|--------|--------|-----|
| Word count | X | Y | +/- Z |
| [other metrics] | | | |

### Issues Detected
1. [Issue 1 — from feedback + diagnostic]
2. [Issue 2]

### Recommended Entry Point
- [ ] Complete restructure (Phase X of skill)
- [ ] Targeted revision (Phase Y of skill)
- [ ] Calibration only (Phase Z of skill)
- [ ] Direct revision (no skill needed)

### Notes
[Any additional observations]

Task 3: Synthesize Across Sections

After running diagnostics for all sections, look for patterns:

1. Cluster coherence: Do intro, theory, and conclusion share the same cluster type? (They should!)

• If intro is Gap-Filler but theory is Theory-Extender, there's a framing mismatch

2. Severity distribution: How many sections need restructuring vs. targeted fixes?

• If most sections need restructuring, this is closer to a rewrite than a revision

3. Dependency implications: Do any diagnostics reveal new dependencies?

• E.g., theory cluster changed → intro and conclusion need to match

Task 4: Update the Revision Map

Add diagnostic findings to revision-map.md:

---

## Phase 1 Diagnostics

### Section: Introduction
- **Cluster**: Gap-Filler
- **Key issues**: Two intros (structural), too long (calibration)
- **Entry point**: interview-bookends Phase 1 (redraft intro)

### Section: Theory
- **Cluster**: Gap-Filler (matches intro)
- **Key issues**: Repetitive with intro, turn is weak
- **Entry point**: lit-writeup Phase 5 (revision) with focus on turn

### Section: Methods
- **Pathway**: Standard (should be Detailed given vulnerability)
- **Key issues**: Missing credibility detail, pathway mismatch
- **Entry point**: methods-writer Phase 1 (redraft with Detailed pathway)

[Continue for each section...]

---

## Coherence Check

- Intro cluster: Gap-Filler ✓
- Theory cluster: Gap-Filler ✓
- Conclusion cluster: [to check after theory revised]
- Cross-cluster alignment: OK / MISMATCH

---

## Revised Sequence

Based on diagnostics:
1. Theory (Phase 5) — foundation, needs turn strengthening
2. Methods (Phase 1) — pathway change needed
3. Introduction (Phase 1) — restructure to eliminate repetition
4. [etc.]

Issue Severity Classification

Common Diagnostic Findings

"Two Intros" Pattern

Symptom: Introduction and first section of lit review cover similar ground Diagnosis: Usually a structural issue—no clear division between intro (puzzle + roadmap) and theory (literature engagement) Resolution: Keep ONE intro; convert the other to proper theory section with lit-writeup guidance

"Weak Turn" Pattern

Symptom: Feedback says gap is unclear or contribution is vague Diagnosis: Turn exists but is buried, generic, or too brief Resolution: lit-writeup Phase 4 focus—make turn explicit, position it correctly, sharpen language

"Methods Credibility Gap" Pattern

Symptom: Feedback asks for more detail on coding, reliability, or process Diagnosis: Often a pathway mismatch—using Efficient or Standard when Detailed is warranted Resolution: methods-writer rediagnosis; may need pathway upgrade

"Promise-Delivery Mismatch" Pattern

Symptom: Feedback says intro promises don't match findings Diagnosis: Coherence issue—intro may have been written before findings were finalized Resolution: Decide which is authoritative (usually findings); revise intro to match; check conclusion callbacks

Output Files

Update in revision/ directory:

• revision-map.md — add Phase 1 diagnostics section

Optionally create:

• diagnostics/intro-diagnostic.md
• diagnostics/theory-diagnostic.md
• etc.

When You're Done

Summarize for the orchestrator:

• Diagnostic findings per section
• Entry points recommended
• Any cluster/coherence mismatches discovered
• Revised sequencing if needed

Example summary:

Diagnostics complete. Theory section is Gap-Filler cluster, word count high (+200), turn needs strengthening → lit-writeup Phase 5. Methods currently Standard but should be Detailed given vulnerable population → methods-writer Phase 1 for pathway change. Introduction has two-intros problem → interview-bookends Phase 1 after theory settled. No cluster mismatches detected. Revised sequence: Theory → Methods → Intro → Conclusion.

Then pause for user confirmation before proceeding to Phase 2.

Phase 2: Skill Dispatch

Why This Phase Matters

This is where revision actually happens. Based on the diagnostic assessment, you now invoke specialized skills to revise each section. The key is providing each skill with the right context so it can do its job effectively.

Dispatch Principles

1. One section at a time: Complete each section before moving to the next (unless dependencies require otherwise)

2. Full context provision: Skills need more than just "fix this"—they need the existing text, the specific feedback, and the cluster/pathway information

3. Track as you go: Update the revision map after each section is completed

4. User checkpoints: After each major section, pause for user review

Dispatch Protocols by Skill

Interview-Bookends Dispatch (Introduction or Conclusion)

What to provide the skill:

## Revision Request: [Introduction/Conclusion]

### Skill: interview-bookends
### Phase: [1 for intro, 2 for conclusion, 3 for coherence check]

### Cluster Identified
[Gap-Filler / Theory-Extension / Concept-Building / Synthesis / Problem-Driven]

### Current Section
[Paste the current introduction or conclusion]

### Feedback Items to Address
1. [ID: I-1] [Specific feedback item]
2. [ID: I-2] [Specific feedback item]

### Required Context
**Theory section summary** (for intro) or **full theory section** (for conclusion callbacks):
[Paste or summarize]

**Findings section summary**:
[Key findings that intro should promise and conclusion should reference]

**Current conclusion** (if revising intro, for coherence):
[Paste current conclusion so intro can set up callbacks]

### Constraints
- Word limit: [if any]
- Elements that must be preserved: [if any]
- Journal requirements: [if any]

### Instructions
Revise this [introduction/conclusion] to address the feedback items while maintaining [Cluster] style. Ensure:
- Opening move matches cluster type
- [For intro] Promises align with what findings deliver
- [For conclusion] Callbacks to intro language are present
- Contribution is explicit and specific

---

Lit-Writeup Dispatch (Theory Section)

What to provide the skill:

## Revision Request: Theory Section

### Skill: lit-writeup
### Phase: [4 for turn focus, 5 for full revision]

### Cluster Identified
[Gap-Filler / Theory-Extender / Concept-Builder / Synthesis / Problem-Driven]

### Current Section
[Paste the current theory/literature section]

### Feedback Items to Address
1. [ID: T-1] [Specific feedback item]
2. [ID: T-2] [Specific feedback item]

### Required Context
**Research question**:
[The study's central question]

**Main argument**:
[The paper's core claim/contribution]

**Key literatures engaged**:
1. [Literature 1]
2. [Literature 2]
3. [Literature 3 if applicable]

### Calibration Targets
| Metric | Target | Current |
|--------|--------|---------|
| Word count | 1145-1744 | [current] |
| Citation density | ~24/1000 words | [current] |
| Subsections | 2-4 | [current] |

### Instructions
Revise this theory section to address the feedback items while maintaining [Cluster] style. Ensure:
- Turn (gap → contribution) is explicit and well-positioned
- Each subsection has clear function in the arc
- [Cluster-specific requirement, e.g., "gap is framed as absence, not criticism"]
- Calibration targets are met

---

Methods-Writer Dispatch

What to provide the skill:

## Revision Request: Methods Section

### Skill: methods-writer
### Phase: [1 for pathway change, 2 for revision within pathway]

### Pathway Identified
[Efficient / Standard / Detailed]

### Current Section
[Paste the current methods section]

### Feedback Items to Address
1. [ID: M-1] [Specific feedback item]
2. [ID: M-2] [Specific feedback item]

### Study Details
- Sample size: [N]
- Population: [description]
- Recruitment: [approach]
- Interview format: [semi-structured, etc.] / Duration: [range]
- Analysis approach: [method]

### Pathway Triggers
- Vulnerable population? [Yes/No — if Yes, triggers Detailed]
- Novel methods? [Yes/No — if Yes, triggers Detailed]
- Space constraints? [Yes/No — if Yes, may warrant Efficient]

### Calibration Targets
| Metric | Target (for pathway) |
|--------|---------------------|
| Word count | [pathway target] |
| Table | [expected?] |
| Positionality | [triggered?] |

### Instructions
Revise this methods section following the [Pathway] template. Address feedback items while ensuring:
- All required components are present (sampling, access, analysis, limitations)
- Saturation/sufficiency is addressed
- [If positionality triggered] Positionality statement included
- Word count is within pathway range

---

Case-Justification Dispatch

What to provide the skill:

## Revision Request: Case Justification

### Skill: case-justification
### Phase: [1 for cluster change, 2 for revision within cluster]

### Cluster Identified
[Minimal / Standard / Deep Historical / Comparative / Policy-Driven]

### Position Check
- Current position: [Before/After theory section]
- Required position: [Before/After theory section]
- Position change needed? [Yes/No]

### Current Section
[Paste the current case justification, if exists]

### Feedback Items to Address
1. [ID: C-1] [Specific feedback item]
2. [ID: C-2] [Specific feedback item]

### Case Details
- Site(s): [name/description]
- Population: [description]
- Key contextual features: [list]
- Relationship to theory: [why this case for this argument?]

### Calibration Targets
| Metric | Target (for cluster) |
|--------|---------------------|
| Word count | [cluster target] |
| Opening move | [cluster expectation] |

### Instructions
Revise this case justification following the [Cluster] template. Address feedback items while ensuring:
- Position relative to theory section is correct
- Opening move matches cluster type
- Justification connects case features to theoretical leverage
- Word count is within cluster range

---

Direct Revision Dispatch (Findings, Discussion)

For sections without dedicated skills, provide structured guidance:

## Revision Request: [Findings/Discussion]

### Approach: Direct revision with guidance

### Current Section
[Paste the current section]

### Feedback Items to Address
1. [ID: F-1] [Specific feedback item]
2. [ID: F-2] [Specific feedback item]

### Revision Guidelines

**For Findings**:
- Organization: By theme/concept (not by interview or chronology)
- Evidence: Each claim supported by quote or count
- Interpretation: Evidence → interpretation rhythm maintained
- Theoretical language: Concepts from theory section appear
- Balance: Subsections roughly proportional (unless justified)

**For Discussion**:
- Summary: Brief recap of main findings (not repetition)
- Implications: How findings advance theory/literature
- Scope conditions: Explicit boundaries on claims
- Limitations: Honest but not self-undermining
- Future directions: 2-3 specific suggestions (optional)

### Context
**From theory section** (concepts that should appear):
[List key theoretical concepts]

**From findings** (claims discussion should interpret):
[List main findings]

### Instructions
Revise this section to address the feedback items. Maintain:
- Clear topic sentences for each paragraph
- Evidence-interpretation rhythm
- Appropriate hedging for claims

Dispatch Sequencing

Follow the sequence established in Phase 1, respecting dependencies:

Typical Sequence

1. Theory section first (if changing) — This is the foundation 2. Methods (usually independent) — Can often run parallel to theory 3. Case justification (if needed) — May depend on theory framing 4. Findings (if changing) — May need terminology updates from theory 5. Introduction — After theory and findings settled 6. Discussion — After findings settled 7. Conclusion — Last, after intro finalized

Dependency Handling

If dependencies exist:

• Complete the upstream section first
• Note what changed that affects downstream sections
• Provide changed context to downstream skills

Example: If theory framing changed from Gap-Filler to Theory-Extension: 1. Complete theory revision 2. Before dispatching intro, note: "Theory cluster changed to Theory-Extension; intro should match" 3. When dispatching conclusion, include revised intro for callback alignment

Tracking Progress

After each dispatch, update revision-map.md:

### Dispatch Log

| Time | Section | Skill/Phase | Items Addressed | Status |
|------|---------|-------------|-----------------|--------|
| [timestamp] | Theory | lit-writeup/P5 | T-1, T-2, T-3 | Complete |
| [timestamp] | Methods | methods-writer/P1 | M-1, M-2 | Complete |
| [timestamp] | Introduction | interview-bookends/P1 | I-1, I-2 | In progress |

### Item Status Update

| ID | Original Status | Current Status | Notes |
|----|-----------------|----------------|-------|
| T-1 | [ ] | [x] | Word count reduced by 300 |
| T-2 | [ ] | [x] | Turn sharpened, moved to end of section |
| T-3 | [ ] | [x] | "Organizational absorption" defined in para 2 |
| M-1 | [ ] | [x] | Coding process expanded |
| M-2 | [ ] | [x] | Pathway changed to Detailed |
| I-1 | [ ] | [~] | In progress |
| I-2 | [ ] | [~] | In progress |

User Checkpoints

After completing each major section, pause:

Section Complete: [Theory Section]

>

Changes made:

- [Summary of changes]

>

Feedback items addressed:

- T-1: [how addressed]

- T-2: [how addressed]

>

Next section: [Introduction]

>

Ready to proceed?

Handling Problems

Skill Can't Fully Address Feedback

If a skill's revision doesn't fully address a feedback item:

• Document what was addressed vs. what remains
• Flag item as [~] (partially addressed) or [!] (needs user decision)
• Note the gap for Phase 4 verification

Contradictory Feedback

If R1 and R2 give contradictory feedback:

• Do NOT resolve this yourself
• Flag as [!] with both options documented
• Ask user which direction to take
• Proceed with user's choice

Scope Creep

If addressing feedback reveals additional issues:

• Add new items to the revision map
• Flag as "emerged during revision"
• Prioritize: address if directly related, defer if tangential

Output Files

For each dispatched section, save:

• section-revisions/[section]-revised.md — The revised section text
• Update revision-map.md — Tracking updates

When You're Done

After all sections have been dispatched and revised:

Summarize for the orchestrator:

• Sections revised
• Items addressed per section
• Any items requiring user decision ([!])
• Any dependencies that need Phase 3 attention

Example summary:

Dispatch complete. 6 sections revised:

- Theory: 3/3 items addressed

- Methods: 2/2 items addressed (pathway changed to Detailed)

- Introduction: 2/2 items addressed

- Findings: 1/1 item addressed

- Discussion: 2/2 items addressed

- Conclusion: 1/1 item addressed

>

All items marked complete except D-1 (scope conditions) flagged [!] — user needs to confirm which scope conditions are appropriate.

>

Ready for Phase 3 integration review.

Then proceed to Phase 3.

Phase 3: Integration Review

Why This Phase Matters

Revising sections individually can introduce incoherence. A strengthened theory section may now use different language than the findings. A restructured intro may no longer set up the conclusion's callbacks. This phase ensures the revised manuscript works as a whole.

Integration Checks

Check 1: Cluster Coherence

Do intro, theory, and conclusion share the same cluster type?

If mismatch detected:

• Usually the theory cluster is authoritative (it reflects the actual contribution)
• Intro and conclusion should be adjusted to match theory
• Use interview-bookends Phase 3 (coherence) to realign

Check 2: Promise-Delivery Alignment

Does the intro promise what the findings deliver?

Extract from intro:

• What puzzle/question does intro pose?
• What is the promised contribution?
• What empirical context is set up?

Extract from findings:

• What does the paper actually deliver?
• What claims are actually made?
• What empirical context is actually covered?

Common mismatches:

Check 3: Theory → Findings Language

Do the theoretical concepts from the revised theory section appear in the findings?

Concept inventory:

If concepts missing from findings:

• Either the findings don't actually engage the theory (substantive problem)
• Or the terminology shifted during revision (fixable with word changes)

If findings use concepts not in theory:

• Either theory section needs to introduce them
• Or findings should use theory's terminology instead

Check 4: Conclusion Callbacks

Does the conclusion reference specific language/concepts from the intro?

Callback inventory:

Interview-bookends benchmark: 100% of strong conclusions have callbacks.

If callbacks missing:

• Identify 2-3 specific intro phrases that can be echoed
• Add explicit references in conclusion (not repetition, but resonance)
• Use interview-bookends Phase 3 specifically for this

Check 5: Methods → Findings Alignment

Do the methods described support the claims made in findings?

Check for:

• If findings claim saturation, does methods mention saturation?
• If findings report precise counts, does methods explain counting procedure?
• If findings discuss negative cases, does methods explain how they were identified?

Check 6: Terminology Consistency

Are key terms used consistently throughout?

Build a terminology table:

Common problems:

• Term defined in theory, synonym used in findings
• Term used before it's defined
• Term definition shifts between sections

Check 7: Citation Consistency

Are the same sources cited consistently?

Check for:

• Same author cited differently (Smith 2020 vs. Smith 2019)
• In-text citations that don't appear in references
• References that are never cited in-text

Integration Repair Protocol

For each integration issue detected:

1. Classify severity:

• Minor: Terminology inconsistency, missing callback (easy fix)
• Moderate: Promise-delivery mismatch requiring section adjustment
• Major: Cluster mismatch requiring re-revision

2. Document the issue:

   ### Integration Issue: [Name]

   **Type**: [Cluster / Promise-Delivery / Terminology / Callback / Citation]
   **Severity**: [Minor / Moderate / Major]
   **Sections affected**: [list]

   **Description**:
   [What's inconsistent and where]

   **Proposed resolution**:
   [How to fix]

   **Action**:
   - [ ] Direct fix (can do now)
   - [ ] Requires skill re-dispatch
   - [ ] Requires user decision

3. Make repairs:

• Minor issues: Fix directly
• Moderate issues: May require skill re-dispatch (back to Phase 2)
• Major issues: Flag for user decision

Cross-Section Coherence Matrix

Create a summary matrix:

## Coherence Matrix

| Check | Status | Notes |
|-------|--------|-------|
| Cluster coherence | ✓/✗ | [notes] |
| Promise-delivery | ✓/✗ | [notes] |
| Theory→Findings language | ✓/✗ | [notes] |
| Conclusion callbacks | ✓/✗ | [notes] |
| Methods→Findings | ✓/✗ | [notes] |
| Terminology | ✓/✗ | [notes] |
| Citations | ✓/✗ | [notes] |

### Issues Requiring Attention
1. [Issue 1 — severity, resolution]
2. [Issue 2 — severity, resolution]

Special Case: Cascading Changes

Sometimes fixing one integration issue creates another. For example:

• Fixing promise-delivery mismatch by changing intro → breaks conclusion callbacks
• Fixing terminology in theory → requires updates in findings AND discussion

Handle cascading changes by: 1. Identifying the full cascade before making changes 2. Making changes in dependency order (upstream → downstream) 3. Re-checking after each change

Output Files

Update in revision/ directory:

• revision-map.md — Add integration findings section
• integration-report.md — Detailed coherence analysis (optional)

When You're Done

Summarize for the orchestrator:

• Integration checks passed/failed
• Issues detected and resolved
• Any issues requiring user decision
• Recommendation: ready for Phase 4 or needs more work

Example summary:

Integration review complete. 5/7 checks passed.

>

Issues detected:

1. Conclusion callbacks: Missing reference to intro's "dispersal" framing — FIXED (added callback in conclusion para 2)

2. Terminology: "Organizational absorption" used inconsistently (sometimes "total institution") — FIXED (standardized to "organizational absorption" throughout)

>

Remaining:

- Cluster coherence: ✓

- Promise-delivery: ✓ (intro now matches revised findings emphasis)

- Theory→Findings: ✓

- Methods→Findings: ✓

- Citations: ✓

>

Ready for Phase 4 verification.

Then proceed to Phase 4.

Phase 4: Verification & Response

Why This Phase Matters

Before declaring revision complete, you need to verify that all feedback has actually been addressed. This phase creates accountability—a clear record of what was changed, how each feedback item was handled, and any remaining gaps.

If the revision is for a journal R&R, this phase also produces the response memo.

Verification Tasks

Task 1: Feedback Item Audit

Go through EVERY feedback item from Phase 0 and verify its status:

## Feedback Item Audit

### Introduction Items
| ID | Feedback | Status | How Addressed | Location |
|----|----------|--------|---------------|----------|
| I-1 | Too long | [x] | Cut 200 words, removed redundant para | Intro para 2 removed |
| I-2 | Two intros | [x] | Merged into single intro | N/A |

### Theory Items
| ID | Feedback | Status | How Addressed | Location |
|----|----------|--------|---------------|----------|
| T-1 | Compress 30% | [x] | Cut 450 words | Throughout |
| T-2 | Strengthen turn | [x] | Sharpened gap→contribution pivot | Theory, final para |
| T-3 | Define "absorption" | [x] | Added definition | Theory para 2 |

### Methods Items
| ID | Feedback | Status | How Addressed | Location |
|----|----------|--------|---------------|----------|
| M-1 | Coding detail | [x] | Expanded coding process description | Methods para 4-5 |
| M-2 | Reliability | [x] | Added structured recoding statement | Methods para 5 |

### [Continue for all sections...]

Status codes:

• [x] — Fully addressed
• [~] — Partially addressed (explain what remains)
• [!] — Requires user decision (explain options)
• [-] — Declined to address (explain why)

Task 2: Change Summary

Create a section-by-section summary of what changed:

## Revision Summary

### Introduction
**Word count**: 850 → 650 (-200 words)
**Structural changes**: Merged two introductions into one; removed redundant "Background" section
**Key improvements**: Opening now directly poses the puzzle; roadmap is clearer

### Theory/Literature Review
**Word count**: 1800 → 1350 (-450 words)
**Structural changes**: None (same 3 subsections)
**Key improvements**: Turn (gap → contribution) is now explicit and positioned at section end; "organizational absorption" defined with observable indicators

### Methods
**Word count**: 900 → 1400 (+500 words)
**Structural changes**: Changed from Efficient to Detailed pathway
**Key improvements**: Added coding process (para 4-5), structured recoding (para 5), saturation statement (para 6)

### Findings
**Word count**: 4000 → 3850 (-150 words)
**Structural changes**: None
**Key improvements**: Aligned terminology with revised theory section; tightened some evidence-interpretation sequences

### Discussion
**Word count**: 1200 → 1400 (+200 words)
**Structural changes**: Added scope conditions subsection
**Key improvements**: Explicit scope conditions; stronger connection to literatures

### Conclusion
**Word count**: 600 → 550 (-50 words)
**Structural changes**: None
**Key improvements**: Added callbacks to intro language ("dispersal" framing)

Task 3: Unaddressed Items Documentation

For any items marked [~], [!], or [-], provide detailed explanation:

## Unaddressed or Partially Addressed Items

### [!] D-1: Scope conditions
**Feedback**: "Add scope conditions"
**Status**: Requires user decision
**Options**:
1. Limit to high-intensity movements only (narrower, safer)
2. Propose generalization to any identity-absorbing organization (broader, riskier)
3. Add both as explicit contrasting scope conditions

**Recommendation**: Option 3 (present both, let readers judge)

### [~] F-1: More quantitative data
**Feedback**: "Consider adding more quantitative data"
**Status**: Partially addressed
**What was done**: Added counts for each exit pathway type (Table 2)
**What remains**: Did not add multivariate analysis (beyond scope of qualitative approach)
**Rationale**: Qualitative methodology doesn't support statistical inference; counts are descriptive

### [-] R2-3: Compare to other movements
**Feedback**: "Would be strengthened by comparison to other movements"
**Status**: Declined to address
**Rationale**:
- Would require additional data collection beyond revision scope
- Comparative claims are already hedged as future research direction
- Single-case depth is methodological strength, not weakness

Task 4: Word Count Reconciliation

If word limits are a constraint, verify compliance:

## Word Count Summary

| Section | Original | Revised | Change |
|---------|----------|---------|--------|
| Abstract | 150 | 148 | -2 |
| Introduction | 850 | 650 | -200 |
| Theory | 1800 | 1350 | -450 |
| Methods | 900 | 1400 | +500 |
| Findings | 4000 | 3850 | -150 |
| Discussion | 1200 | 1400 | +200 |
| Conclusion | 600 | 550 | -50 |
| **Total** | **9500** | **9348** | **-152** |

**Target**: 10,000 words
**Status**: ✓ Within limit

Task 5: Response Memo (for R&R)

If this is a journal revision, draft a response memo:

# Response to Reviewers

## Editor
Thank you for the opportunity to revise this manuscript. We have carefully considered all feedback and made substantial revisions as detailed below.

## Reviewer 1

### Comment 1.1
> "The introduction is too long and repetitive—you have two intros."

**Response**: We agree and have restructured the opening. The original manuscript had both an Introduction and a "Background" section that covered similar ground. We have merged these into a single, streamlined Introduction (now 650 words, down from 850). The Introduction now opens directly with the empirical puzzle, presents the gap, previews our contribution, and provides a roadmap—without the repetition the reviewer correctly identified.

**Changes**: See revised Introduction, pp. 1-2.

### Comment 1.2
> "Methods need more detail on coding."

**Response**: We have substantially expanded the Methods section (now 1400 words, up from 900) to address this concern. Specifically, we added:
- A detailed description of our three-stage coding process (para 4)
- Explanation of how we distinguished key constructs (pace exhaustion vs. grief accumulation vs. fit failure) (para 5)
- Statement on structured recoding after refining definitions (para 5)

**Changes**: See revised Methods, pp. X-Y, paragraphs 4-5.

### [Continue for all feedback items...]

## Reviewer 2

[Same format...]

## Summary of Major Changes

1. **Structural**: Merged two introductions; added scope conditions to Discussion
2. **Methodological**: Expanded coding detail; clarified construct distinctions
3. **Conceptual**: Sharpened the turn (gap → contribution); defined "organizational absorption" with observable indicators
4. **Presentational**: Compressed theory section by 25%; added Table 2 (exit pathway counts)

We believe these revisions address the reviewers' concerns while preserving the manuscript's core contributions. Thank you again for the constructive feedback.

Verification Checklist

Before declaring revision complete, confirm:

• [ ] All feedback items have a status ([x], [~], [!], or [-])
• [ ] All [~] items have explanation of what remains
• [ ] All [!] items have options presented to user
• [ ] All [-] items have rationale for declining
• [ ] Change summary covers every section
• [ ] Word count is within limits (if applicable)
• [ ] Response memo drafted (if R&R)
• [ ] Revised manuscript assembled in correct order

Output Files

Create in revision/ directory:

• revision-summary.md — Complete accounting of changes
• response-memo.md — Reviewer response (if R&R)
• revised-draft.md — Complete revised manuscript

Final Assembly

Assemble the revised manuscript in correct order:

1. Title page (if required) 2. Abstract 3. Introduction 4. Theory/Literature Review 5. Case Justification (if Policy-Driven cluster: before Theory) 6. Methods 7. Findings 8. Discussion 9. Conclusion 10. References 11. Tables and Figures 12. Appendices

When You're Done

Summarize for user:

• Total feedback items: X
• Fully addressed: Y
• Partially addressed: Z (with reasons)
• Requires decision: W (with options)
• Declined: V (with rationale)
• Word count change: ±N words
• Response memo: Ready/Not needed

Example final summary:

Revision complete.

>

Feedback accountability:

- 11 feedback items total

- 9 fully addressed [x]

- 1 partially addressed [~] (quantitative data — added counts, declined multivariate)

- 1 requires decision [!] (scope conditions — three options presented)

- 0 declined

>

Word count: 9500 → 9348 (-152 words), within 10K limit

>

Deliverables:

- revised-draft.md — Complete revised manuscript

- revision-summary.md — Full change documentation

- response-memo.md — Reviewer response ready for review

>

Action needed: Please confirm scope conditions choice (D-1) before finalizing.

Revision coordination complete.