Help · FAQ
Short answers on ChunkMaps for humans and agentic systems. Each answer links to a deeper guide.
- What is a Chunk Map?
- A Chunk Map is a structured spatial model of ideas that links chunking up, down, and across inside a Domain. It is the core map for persuasion, clarity, and agentic architecture. Read more →
- What do chunking up, down, and across mean?
- Up raises abstraction to purpose and category. Down adds specifics and evidence. Across finds peer alternatives at the same level. Together they keep maps navigable without losing the thread. Read more →
- Why does ChunkMaps use 7 plus or minus 2?
- Miller's working-memory limit keeps top-level siblings cognitively manageable. ChunkMaps treats that as a design rule for people and for LLM context windows. Read more →
- What is CHUNKS_OK?
- CHUNKS_OK is the persuasion completeness template: Current Situation, Headaches, Utopia, Next Steps, Knowledge, Solution, Offer, and closing Knowledge. It ensures pitches cover the full arc. Read more →
- What are the 4 Chunk Map Models?
- Definition, Interaction, Environment, and Features. Use them in order to place chunks, map interactions, read context, and extract differentiating features. Read more →
- What is the Persuasion Flow?
- A reusable procedure: set Domain and outcome, build a Chunk Map, plot a Journey of Persuasion, map counterarguments, then encode CHUNKS_OK. A guided page and .md skill are available. Read more →
- What is ChunkMaps for the Agentic Age?
- The same hierarchical chunking that aligns people now structures DAG prompt-trees, shared agent state, and enterprise logic so autonomy stays coherent. Start at the Agentic ChunkMaps hub. Read more →
- How do orchestrators stop multi-agent drift?
- Use DAG prompt-trees plus a shared blackboard with write contracts. Each agent node may write only allowed state paths and must cite facts instead of inventing commitments. Read more →
- How does ChunkMaps help enterprise policy and decisions?
- Policy-to-pipeline compiles unstructured policy into versioned, citable logic chunks for automated evaluation. That supports enterprise decision-making frameworks agents can execute. Read more →
- How does ChunkMaps support AI prompt engineering?
- Spatial chunk maps replace flat mega-prompts with multi-dimensional maps that govern context, constrain drift, and structure multi-step reasoning-useful as AI prompt chaining frameworks. Read more →
- What is the skills library?
- Editable .md playbooks (persuasion flow, agent blackboard, policy-to-pipeline) that agents and humans can load step-by-step. Browse the Skills library page, then open how-to and raw .md files as needed. Read more →
- Where is the machine-readable ChunkMaps spec?
- Open taxonomy for developer agents is published as JSON and YAML under spec/. It defines Domain, vectors, four models, CHUNKS_OK, and optional agentic extensions. Read more →