visualiserlisted

# Visualiser **Speaks map format v1.** Reads the **index**. Renders for a **human**. Writes no claims and changes no status. If you find something wrong with the map while rendering it, say so — do not fix it here; that is `cartographer`. ## Why this exists A map that has to be *consulted* through an assistant is not readable. The symptom is a human asking what their own map says. Your output should mean they no longer have to ask. ## 1. Render from the index, never from the claims 🛑 **Read the index.** It is the only part of the map that knows what is current. Rendering from the underlying claims will draw superseded findings beautifully and confidently, which is the worst failure available here. If the map declares a format version other than v1, stop and say so. ## 2. Ask what the human is looking at it for One question before drawing: **what are you trying to see?** The answer picks the view. If they do not know, offer the four: | View | Answers | Shape | |---|---|---| | **Constellation** | What is here and how does it relate? | Node graph of subjects and relations | | **Architecture** | How is this thing put together internally? | Component or sequence diagram | | **Coverage** | Where is the map thin, and where should I survey next? | The territory shaded by resolution | | **Status** | What is settled, contested, or stale? | Board or table by index status | **Coverage is the one people do not think to ask for and most often need** — it is the only view that s