Lit Writeup

Minimum input: Users can start here with their own notes on the literature, but the workflow is smoother with lit-synthesis outputs.

When to Use This Skill

Use this skill when users want to:

• Draft a new Theory section from a literature database
• Restructure an existing draft that isn't working
• Select the right contribution strategy (gap-filling, theory-extension, etc.)
• Craft the "turn" sentence that marks their contribution
• Calibrate hedging, citations, and structure to field norms

Core Principles

1. Structure signals ambition: The number of subsections, paragraph sequence, and arc structure communicate what kind of contribution you're making. Match form to content.

2. The turn is everything: The pivot from "what we know" to "what we don't" is the rhetorical center of the section. Craft it carefully.

3. Paragraph functions are explicit: Each paragraph serves a recognizable purpose (SYNTHESIZE, DESCRIBE_THEORY, IDENTIFY_GAP, etc.). Readers should sense the function even without subheadings.

4. Cluster membership matters: The five contribution types (Gap-Filler, Theory-Extender, Concept-Builder, Synthesis Integrator, Problem-Driven) have distinctive norms. Know which you're writing.

5. Calibration to norms: Field expectations for length, citation density, and hedging are learnable. Deviation should be intentional, not accidental.

The Five Clusters

Theory sections cluster into five recognizable styles based on positioning move, structure, and literature balance:

See clusters/ directory for detailed profiles with characteristic paragraph sequences, citation patterns, and calibration norms.

Workflow Phases

Phase 0: Assessment

Goal: Identify contribution type and select cluster.

Process:

• Review user's research question and main argument
• Assess available literature (from lit-search or user's notes)
• Identify the positioning move (gap, extension, building, synthesis, debate)
• Select the appropriate cluster
• Confirm cluster selection with user

Output: Cluster selection memo with rationale.

Pause: User confirms cluster selection before architecture.

Phase 1: Architecture

Goal: Design section structure, subsections, and arc.

Process:

• Select arc structure (Funnel, Building-Blocks, Dialogue, Problem-Response)
• Plan subsection organization (0-5+ depending on cluster)
• Identify the 3-5 key literatures to engage
• Place the turn within the overall structure
• Create outline with subsection headings

Output: Architecture memo with section outline.

Pause: User approves structure before paragraph planning.

Phase 2: Planning

Goal: Map paragraph functions and sequence.

Process:

• Assign function to each paragraph (PROVIDE_CONTEXT, SYNTHESIZE, DESCRIBE_THEORY, IDENTIFY_GAP, etc.)
• Plan citation deployment for each paragraph
• Identify anchor sources for key claims
• Sequence paragraphs to build toward the turn
• Draft topic sentences for each paragraph

Output: Paragraph map with functions and topic sentences.

Pause: User reviews paragraph map.

Phase 3: Drafting

Goal: Write paragraphs with sentence-level craft.

Process:

• Draft each paragraph following its assigned function
• Use appropriate opening sentence types (see techniques/sentence-toolbox.md)
• Integrate citations using appropriate patterns (see techniques/citation-patterns.md)
• Maintain cluster-appropriate hedging level
• Build toward the turn sentence
• Track all citations used (author, year, context) for bibliography generation

Output: Full draft of Theory section + citations-tracking.json.

Pause: User reviews each subsection (if multiple) or full draft.

Phase 4: Turn

Goal: Craft the gap/contribution pivot.

Process:

• Apply the 4-part turn formula (see techniques/turn-formula.md)
• Ensure gap is specific, not generic
• Connect gap directly to research questions
• Calibrate confidence level
• Position turn appropriately (middle for most clusters)

Output: Refined turn sentence(s) and surrounding context.

Pause: User evaluates the turn for clarity and specificity.

Phase 5: Revision

Goal: Calibrate against norms and polish.

Process:

• Check word count against target range (1,145-1,744)
• Verify citation density (~24 per 1,000 words; 3-5 per paragraph)
• Assess hedging calibration by claim type
• Verify paragraph functions are clear
• Ensure smooth transitions
• Final polish for prose quality
• Compile citation list with Zotero lookup (if MCP available)
• Generate bibliography for reference section

Output: Final Theory section + quality memo + citations-final.json + bibliography.md.

Technique Guides

The skill includes detailed reference guides in techniques/:

Cluster Profiles

Detailed profiles in clusters/:

Calibration Benchmarks

Based on 80 articles from Social Problems and Social Forces:

Invoking Phase Agents

Use the Task tool for each phase:

Task: Phase 0 Assessment
subagent_type: general-purpose