ChakataStat
Download

Settings

Open settings with File → Settings… (Ctrl+,). Settings are remembered between launches. They fall into four areas.

Appearance

  • Theme mode: Light, Dark or follow the system.
  • Accent colour: pick from a curated palette; it re-themes the whole app live.
  • Text size: scale the UI text up or down for readability: a Small / Default / Large preset, or a Custom percentage you type in.
  • Content text size: scale just the reading surfaces (the Output log, the syntax console's messages and the AI chat) on top of the text size, so results and error messages can be enlarged without inflating the whole UI. Presets up to Largest, or a Custom percentage.
  • Chart colours: which validated palette multi-series charts use to tell series apart: Standard (the default) or Colour-blind safe (Okabe–Ito). The list is short and closed on purpose: every entry is checked for colour-vision separation in both light and dark mode, so a palette that would leave two groups indistinguishable is not offered. The swatch strip under the dropdown shows the palette as it will actually paint. See Graphs → Choosing a palette.
  • Output chart width: how wide charts are drawn in the Output log.

Files

  • Protect CSV exports from formulas: on by default. When you export to CSV/TSV, any text cell that starts with =, +, -, @, a tab or a carriage return and is not a number is written with a leading apostrophe ('), which spreadsheets read as "text" and hide, so a value that arrived in your data can never run as a formula in someone's Excel or Calc. Numbers such as -5 are never touched. Turn it off for a byte-faithful export.
  • Mark CSV exports as UTF-8 for Excel: on by default. When a CSV/TSV export holds any non-ASCII text (é, ü, ß, …), the file starts with a byte-order mark, so Excel on Windows reads it as UTF-8 instead of garbling those characters. A plain-ASCII export never gets one. Turn it off if a tool shows a stray character before the first column name (base R's read.csv does unless given fileEncoding = "UTF-8-BOM"; pandas, readr and data.table handle the mark).

Connectivity

A connectivity check used to confirm the app can reach the network (relevant to the AI assistant when using a hosted backend). A local Ollama backend needs no outbound connection.

AI Assistant

Configure the model the assistant talks to:

Controls are grouped into Connection (the active backend and a reference of the environment variables that activate each one, marked set or unset), Behavior (the everyday controls), and a collapsed Connection overrides group for the rarely-needed fields.

  • Backend: choose among the backends you have configured (Claude, an OpenAI-compatible endpoint, DeepSeek, Kimi, Ollama), and switch live. Your choice is persisted; your API key is never stored.
  • Environment variables: the exact variable name for each backend (ANTHROPIC_API_KEY, OPENAI_API_KEY, DEEPSEEK_API_KEY, MOONSHOT_API_KEY, or ChakataStat_OLLAMA=1), each marked set or unset, so you can confirm you used the name the app reads.
  • Overrides: model id, temperature, maximum response tokens, base URL, custom headers, the tool-round cap, a system-prompt override, and the read-only toggle (on by default; the same switch sits in the assistant panel's header). The base URL and custom headers are stored in plain text in your session file; keep API keys in their environment variable, never in a header.
  • Custom context: a free-text note about your data and goals, sent with each request so the assistant interprets your dataset correctly.

Advanced

  • Advanced mode: off by default. Turning it on reveals Engine Diagnostics. It changes nothing about your data; it only exposes the developer-facing tools.

Where settings live

Settings are part of your persisted session, alongside the window layout, the recent-files list, the AI conversation and the syntax script. No API keys are ever written to disk, but the assistant's custom base URL and headers are, in plain text, so never put a key in a header.