chain-of-density

---
name: chain-of-density
description: "Iteratively densify text summaries using Chain-of-Density technique. Use when compressing verbose documentation, condensing requirements, or creating executive summaries while preserving information density."
license: Apache-2.0
compatibility: "Python 3.10+ (for text_metrics.py script via uv run)"
metadata:
  author: agentic-insights
  version: "1.2"
  paper: "From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting"
  arxiv: "https://arxiv.org/abs/2309.04269"
---

# Chain-of-Density Summarization

Compress text through iterative entity injection following the CoD paper methodology. Each pass identifies missing entities from the source and incorporates them while maintaining identical length.

## The Method

Chain-of-Density works through multiple iterations:

1. **Iteration 1**: Create sparse, verbose base summary (4-5 sentences at `target_words`)
2. **Subsequent iterations**: Each iteration:
   - Identify 1-3 missing entities from SOURCE (not summary)
   - Rewrite summary to include them
   - Maintain IDENTICAL word count through compression

**Key principle**: Never drop entities - only add and compress.

## Missing Entity Criteria

Each entity added must meet ALL 5 criteria:

| Criterion | Description |
|-----------|-------------|
| **Relevant** | To the main story/topic |
| **Specific** | Descriptive yet concise (≤5 words) |
| **Novel** | Not in the previous summary |
| **Faithful** | Present in the source (no hallucination) |
| **Anywhere** | Can be from anywhere in the source |

## Quick Start

1. User provides text to summarize
2. Orchestrate 5 iterations via `cod-iteration` agent
3. Each iteration reports entities added via `Missing_Entities:` line
4. Return final summary + entity accumulation history

## Orchestration Pattern

```
Iteration 1: Sparse base (target_words, verbose filler)
     ↓ Missing_Entities: (none - establishing base)
Iteration 2: +3 entities, compress filler
     ↓ Missing_Entiti