Content Iteration

XunLong's iteration system allows you to refine and modify generated content without starting from scratch.

Overview

The Iteration feature:

• 🔄 Modifies existing content intelligently
• 💾 Preserves context and style
• 📝 Targets specific sections or global changes
• 🗂️ Maintains version history
• ⚡ Faster than regeneration

Quick Start

# Generate initial content
python xunlong.py report "AI Trends" --depth standard

# Iterate on it
python xunlong.py iterate <project-id> "Add more examples in the conclusion"

Modification Scopes

Local Scope 📍

Target: Single section, paragraph, or chapter

Use cases:

• Fix typos
• Update specific data
• Rewrite a paragraph
• Add/remove a sentence

Example:

python xunlong.py iterate <project-id> \
  "Fix the typo in Chapter 3, paragraph 2"

Process:

Speed: ~30 seconds

Partial Scope 🎯

Target: Multiple sections or chapters

Use cases:

• Add new sections
• Reorganize chapters
• Expand specific topics
• Remove redundant parts

Example:

python xunlong.py iterate <project-id> \
  "Add three case studies to chapters 4, 5, and 6"

Process:

Speed: ~2-5 minutes

Global Scope 🌐

Target: Entire document

Use cases:

• Change overall tone
• Add theme throughout
• Restructure document
• Change style/format

Example:

python xunlong.py iterate <project-id> \
  "Make the entire report more technical and add code examples throughout"

Process:

Speed: ~5-15 minutes

Common Iteration Tasks

Adding Content ➕

# Add new section
python xunlong.py iterate <project-id> \
  "Add a new section on 'Future Trends' after the current conclusion"

# Add examples
python xunlong.py iterate <project-id> \
  "Add 2-3 real-world examples to the Implementation section"

# Add data
python xunlong.py iterate <project-id> \
  "Include recent statistics and market data in Chapter 2"

# Add citations
python xunlong.py iterate <project-id> \
  "Add more academic citations to support claims in the Literature Review"

Removing Content ➖

# Remove section
python xunlong.py iterate <project-id> \
  "Remove the Technical Details section as it's too in-depth"

# Trim content
python xunlong.py iterate <project-id> \
  "Shorten Chapter 5 by about 30%, focusing on key points only"

# Remove redundancy
python xunlong.py iterate <project-id> \
  "Remove redundant information between chapters 3 and 7"

Modifying Content ✏️

# Change tone
python xunlong.py iterate <project-id> \
  "Make Chapter 4 more casual and conversational"

# Improve clarity
python xunlong.py iterate <project-id> \
  "Simplify the technical explanations in the Methodology section"

# Enhance detail
python xunlong.py iterate <project-id> \
  "Expand the character development in chapters 8-10"

# Update information
python xunlong.py iterate <project-id> \
  "Update the market data with 2025 figures"

Restructuring Content 🔀

# Reorder sections
python xunlong.py iterate <project-id> \
  "Move the Case Studies section before the Analysis section"

# Split sections
python xunlong.py iterate <project-id> \
  "Split Chapter 6 into two separate chapters"

# Merge sections
python xunlong.py iterate <project-id> \
  "Combine chapters 2 and 3 into a single Overview chapter"

Style Changes 🎨

# Change writing style
python xunlong.py iterate <project-id> \
  "Convert the academic tone to a more business-friendly style"

# Adjust formality
python xunlong.py iterate <project-id> \
  "Make the conclusion more formal and authoritative"

# Change perspective
python xunlong.py iterate <project-id> \
  "Rewrite chapter 5 from first-person to third-person perspective"

Version Management

Automatic Versioning

Every iteration creates a new version:

storage/project_id/
├── versions/
│   ├── v1_20251005_143022/  # Original
│   ├── v2_20251005_150130/  # After 1st iteration
│   ├── v3_20251005_152045/  # After 2nd iteration
│   └── v4_20251005_154512/  # Current
└── reports/
    └── FINAL_REPORT.md      # Latest version

View Version History

python xunlong.py versions <project-id>

Output:

📚 Version History for project_id

v4 (current) - 2025-10-05 15:45:12
   Modification: "Add more technical details to implementation"
   Changes: +347 words, 3 sections modified

v3 - 2025-10-05 15:20:45
   Modification: "Expand case studies section"
   Changes: +892 words, 1 section added

v2 - 2025-10-05 15:01:30
   Modification: "Fix typos and improve clarity"
   Changes: ±124 words, 8 sections modified

v1 - 2025-10-05 14:30:22
   Initial generation

Compare Versions

python xunlong.py diff <project-id> v2 v4

Shows differences between versions.

Rollback to Previous Version

python xunlong.py rollback <project-id> v2

Restores document to version 2.

Advanced Features

Targeted Search and Replace

python xunlong.py iterate <project-id> \
  --mode search-replace \
  --find "machine learning" \
  --replace "artificial intelligence" \
  --scope all

Batch Iterations

Define multiple changes in a file:

# iterations.txt
Add case study to chapter 3
Expand technical details in chapter 5
Add conclusion to chapter 7
Update statistics throughout

python xunlong.py iterate <project-id> --batch iterations.txt

Conditional Iterations

Only modify if condition is met:

python xunlong.py iterate <project-id> \
  "Add COVID-19 context if discussing 2020-2025 trends"

Style Transfer

Apply style from another document:

python xunlong.py iterate <project-id> \
  --apply-style-from reference_document.md

Iteration Workflow

Interactive Mode

python xunlong.py iterate <project-id> --interactive

Process:

1. View current content
2. Specify modification
3. Preview changes
4. Approve or reject
5. Apply if approved

Review Mode

python xunlong.py iterate <project-id> \
  "Add examples" \
  --preview-only

Shows what would change without applying.

Auto-approve Mode

python xunlong.py iterate <project-id> \
  "Fix grammar and spelling" \
  --auto-approve

Applies changes without confirmation.

Quality Assurance

Coherence Checking

After iteration, XunLong checks:

• Logical flow maintained
• No contradictions introduced
• Style consistency preserved
• Transitions smooth

Automatic Cleanup

• Removes duplicate content
• Fixes broken references
• Updates table of contents
• Renumbers sections

Validation

python xunlong.py validate <project-id>

Checks:

• Structure integrity
• Citation validity
• Cross-references
• Format compliance

Performance Optimization

Iteration Speed Comparison

Scope	Speed	Use Case
Local	30 sec	Typos, small edits
Partial	2-5 min	Multiple sections
Global	5-15 min	Major changes
Full Regen	10-30 min	Complete rewrite

Speed Tip: Prefer targeted iterations over full regeneration.

Caching

XunLong caches:

• Document embeddings
• Search results
• Generated sections

Subsequent iterations are faster due to caching.

Parallel Processing

For batch iterations:

python xunlong.py iterate <project-id> \
  --batch changes.txt \
  --parallel

Processes independent changes simultaneously.

Best Practices

📝 Clear Instructions

Good:

• "Add a table comparing features in the Overview section"
• "Rewrite the introduction to be more engaging"
• "Expand chapter 3 with 2-3 real-world examples"

Less Effective:

• "Make it better" (too vague)
• "Fix everything" (unclear scope)
• "Change stuff" (no direction)

🎯 Iterative Approach

Recommended workflow:

# Step 1: Generate base content
python xunlong.py report "Topic"

# Step 2: Review
cat storage/<project-id>/reports/FINAL_REPORT.md

# Step 3: First iteration (structure)
python xunlong.py iterate <project-id> "Add section on challenges"

# Step 4: Second iteration (content)
python xunlong.py iterate <project-id> "Expand examples in new section"

# Step 5: Final polish
python xunlong.py iterate <project-id> "Improve clarity and fix typos"

⚡ Scope Management

Use local scope when:

• Fixing specific errors
• Updating data points
• Minor adjustments

Use partial scope when:

• Adding/removing sections
• Reorganizing content
• Expanding specific areas

Use global scope when:

• Changing overall style
• Major restructuring
• Adding themes throughout

Troubleshooting

Issue: Changes not applied correctly

Solutions:

# Be more specific
python xunlong.py iterate <project-id> \
  "In Chapter 3, Section 2, add a paragraph about market trends"

# Preview first
python xunlong.py iterate <project-id> \
  "..." \
  --preview-only

Issue: Lost content after iteration

Solutions:

# Rollback to previous version
python xunlong.py rollback <project-id> v3

# View version history
python xunlong.py versions <project-id>

Issue: Iteration too slow

Solutions:

# Use local scope
python xunlong.py iterate <project-id> \
  "Fix typo in paragraph 3" \
  --scope local

# Reduce model size
--model gpt-4o-mini

Issue: Style inconsistency after multiple iterations

Solutions:

# Run style harmonization
python xunlong.py harmonize <project-id>

# Reset and regenerate
python xunlong.py regenerate <project-id> \
  --preserve-content \
  --harmonize-style

API Reference

python xunlong.py iterate <project-id> <instruction> [options]

Argument	Type	Default	Description
<project-id>	str	Required	Project identifier
<instruction>	str	Required	Modification instruction
--scope	str	auto	Modification scope (local/partial/global/auto)
--preview-only	flag	false	Preview changes without applying
--auto-approve	flag	false	Apply without confirmation
--interactive	flag	false	Interactive iteration mode
--batch	str	None	Batch iteration file
--mode	str	intelligent	Iteration mode
--preserve-style	flag	true	Maintain original style
--model	str	gpt-4o-mini	LLM model to use

Examples

Report Iteration

# Initial generation
python xunlong.py report "Cloud Computing Trends 2025"

# Add content
python xunlong.py iterate <project-id> \
  "Add a section on serverless computing trends"

# Expand existing
python xunlong.py iterate <project-id> \
  "Expand the security section with specific examples"

# Update data
python xunlong.py iterate <project-id> \
  "Update market size figures with latest 2025 data"

Fiction Iteration

# Initial generation
python xunlong.py fiction "Mystery novel" --chapters 20

# Character development
python xunlong.py iterate <project-id> \
  "Make the protagonist more conflicted in chapters 5-8"

# Plot enhancement
python xunlong.py iterate <project-id> \
  "Add foreshadowing for the twist in chapters 3, 7, and 11"

# Pacing adjustment
python xunlong.py iterate <project-id> \
  "Slow down the pacing in chapter 15 to build more tension"

PPT Iteration

# Initial generation
python xunlong.py ppt "Product Launch Strategy" --slides 18

# Add slides
python xunlong.py iterate <project-id> \
  "Add 2 slides on competitive analysis after slide 8"

# Improve content
python xunlong.py iterate <project-id> \
  "Make slide 5 more visual with less text"

# Reorder
python xunlong.py iterate <project-id> \
  "Move the pricing slide to come before the features slide"

Next Steps

• Learn about Report Generation
• Explore Fiction Writing
• Try PPT Creation
• Check Export Formats