# ChunkMaps Spec

> Source: [https://chunkmaps.com/NLP_improved_communication/chunkmaps_spec.html](https://chunkmaps.com/NLP_improved_communication/chunkmaps_spec.html)

Open ChunkMaps taxonomy in JSON and YAML for developer agents-Domain, chunking vectors, four models, CHUNKS_OK, plus optional blackboard, DAG, and policy-to-pipeline extensions.

Open standard · Developers

ChunkMaps publishes a machine-readable taxonomy so developer agents and tools can ingest the framework without scraping prose. The same content model ships as JSON and YAML.

## Download the spec

  [chunkmaps.json](spec/chunkmaps.json)
  [chunkmaps.yaml](spec/chunkmaps.yaml)
  [llms.txt](https://chunkmaps.com/llms.txt)
  [FAQ](faq.html)

Canonical public URLs (when live):

- `https://chunkmaps.com/NLP_improved_communication/spec/chunkmaps.json`

- `https://chunkmaps.com/NLP_improved_communication/spec/chunkmaps.yaml`

- `https://chunkmaps.com/llms.txt`

- `https://chunkmaps.com/ai.txt`

## What the taxonomy defines

- **Domain** and desired outcome boundaries

- **Up / Down / Across** vectors with typical questions

- **Working memory** rule (7±2) for cognitive load management

- **Four models:** Definition, Interaction, Environment, Features

- **CHUNKS_OK** persuasion variables

- **Optional agentic extensions:** blackboard, DAG prompt-trees, policy-to-pipeline, spatial reasoning

Use this as an AI prompt chaining and structured argument-mapping schema for LLMs, while humans still learn the method through the Learn path and Agentic hub. Agents can start from [llms.txt](https://chunkmaps.com/llms.txt) for a curated link map.

### Related

- [llms.txt](https://chunkmaps.com/llms.txt) (agent entry map)

- [Agentic ChunkMaps hub](agentic_chunkmaps.html)

- [How to use .md skill files](how_to_use_skill_files.html)

- [What Is A Chunk Map](what_is_a_chunk_map.html)

- [The 4 Chunk Map Models](the_4_chunk_map_models.html)

- [CHUNKS_OK template](best_persuasive_template.html)