stylechart

Describe a mood in one ordinary sentence and stylechart tells you which visual styles it points at. Under the hood is a two-layer lexicon: feeling words on one side, visual styles on the other, and links between them. One colloquial sentence goes in; the best-matching styles come back, together with the feeling words that carried the match, so you can see why a style was chosen rather than just its name.

stylechart is a direct client of the documented public feeling-search REST API (v2) hosted at https://xiximm.cloud. The lexicon and the matching live server-side; this package validates arguments, calls the free endpoints, and hands the JSON back.

The five tools:

• word_analyze — one sentence (up to 500 characters) → the best-matching visual styles plus the feeling words behind them. Optional words (1–12) and styles (1–30) cap how many of each come back. A phrasing outside the lexicon returns an empty word list; that is expected, not an error.

• word_lookup — a single feeling word in full: definition, near-synonyms, and the styles it lands on.

• word_list — the whole lexicon in one response. With a query, the dictionary is filtered locally: words whose text contains the query and styles whose name or id does, with filtered and total counts reported side by side.

• style_get — one style by id (for example synthwave): definition, parent style, member words, reference images.

• word_status — lexicon coverage only: the {ok, counts} slice of the dictionary response, for checking the data is fresh without downloading the full tables.

Free endpoints only

This rebuild talks exclusively to the documented free endpoints (POST /v2/analyze, GET /v2/links, GET /v2/dictionary, GET /v2/styles/<id>). The paid cloud path is not supported: there is no payment machinery of any kind, no support for purchased quota tokens (the X-FS-Quota header is never sent), and the metered images endpoint is never called. If the upstream ever answers /v2/analyze with HTTP 402, stylechart reports it as an error naming the paid path instead of paying.

Related MCP server: Soul Crystal MCP

Configuration

• FEELING_API_BASE — API base URL. Defaults to https://xiximm.cloud. No API key is needed or accepted.

CLI

Each tool is a subcommand taking one JSON argument string (word_status takes none, word_list's argument is optional):

npx @reneturcios/stylechart word_analyze '{"text":"rainy neon streets at night","styles":5}'
npx @reneturcios/stylechart word_lookup '{"word":"melancholy"}'
npx @reneturcios/stylechart word_list '{"query":"neon"}'
npx @reneturcios/stylechart word_list
npx @reneturcios/stylechart style_get '{"id":"synthwave"}'
npx @reneturcios/stylechart word_status
npx @reneturcios/stylechart --version

Output is the tool's result JSON. Exit codes: 0 on success, 1 on a runtime error (including a validation or upstream failure), 2 on a usage error.

MCP server

npx -y @reneturcios/stylechart mcp

MCP client config:

{
  "mcpServers": {
    "stylechart": {
      "command": "npx",
      "args": ["-y", "@reneturcios/stylechart", "mcp"]
    }
  }
}

Docker

docker run -i ghcr.io/cloudcorprecords/stylechart
docker run --rm ghcr.io/cloudcorprecords/stylechart word_status

Development

npm install
npm run build   # tsc -> dist/
npm test        # tsx --test test/*.test.ts (fake fetch only, no network)
npm run dev     # run the CLI from source

License

MIT — Copyright (c) 2026 Rene Turcios