D3Codex
Developers

API & llms.txt

D3Codex is machine-readable. Every topic is exposed as structured JSON with a clean Markdown rendering inside, and an llms.txt discovery file maps the whole reference for LLMs and tools. It is static and CORS-enabled, so you can fetch it from anywhere with no key.

Use it with your AI

One step. Add this single URL to your AI assistant as a documentation / context source. It can then pull any D3 topic on demand, and, in a coding agent (Cursor, Claude Code, Continue), check your D3 / PickBASIC code for errors and open it in the Annotator, using the same engine as the on-site Checker.

https://d3codex.com/llms.txt

In Cursor: Settings → Docs → Add new doc. In Continue or most agents: add the URL as a docs / context source.

Prefer to paste a single page into a chat? Add .md to any topic URL. Building a tool? Use the JSON API below (static, CORS-enabled).

Using Claude Code? Install the skill

One command installs a skill so Claude automatically looks up D3 terms, checks your code with the linter, and opens it in the Annotator, with no URL to paste each time.

mkdir -p ~/.claude/skills/d3codex && curl -s https://d3codex.com/d3codex.skill.md -o ~/.claude/skills/d3codex/SKILL.md
⚡ Try in a prompt

  1. 1 Pick an example below and copy it.

    Using https://d3codex.com/pickbasic-flashbasic/oconv-function.md, write a D3 BASIC program that prints today’s date formatted as MM/DD/YYYY.
    Using https://d3codex.com/api/topics.json as a reference, write a PickBASIC program that asks the user for a date and outputs what day of the week it is (for example: Friday).
    Using https://d3codex.com/llms.txt, check my D3 / PickBASIC code for errors and explain each finding (paste your program after this line).

  2. 2 Open Pick Developer Pro and paste it into the chat.

    ⚡
    Pick Developer Pro ChatGPT
    A custom GPT tuned for D3 / Pick.
    Open →

Open files from your terminal

A small, open shell script, d3codex, opens any local file in the Annotator full-screen straight from your terminal. It encodes the file into the URL hash (nothing is uploaded) and opens your browser. Install it with one line:

curl -fsSL https://d3codex.com/install.sh | bash

The installer puts d3codex in ~/.local/bin and adds it to your PATH if needed (no sudo). Then d3codex myprogram.bp. Works on macOS and Linux (and WSL on Windows); native Windows falls back to printing the URL. Full details, manual install, and the source.

Endpoints

/llms.txt Curated Markdown map of the whole reference (llmstxt.org), for LLMs and tools. → /api/index.json Self-describing API descriptor: endpoints, counts, and an example. → /api/topics.json Catalog of all 2190 topics: slug, title, summary, and links. → /api/topics/{slug}.json One topic: structured sections plus a full Markdown rendering. Slug is "{category}/{topic}". → /{slug}.md Any topic as raw Markdown - just append .md to its page URL. → /api/categories.json The 18 categories with counts. →

Quick start

# every topic is available as JSON (with clean Markdown inside)
curl https://d3codex.com/api/topics/pickbasic-flashbasic/oconv-function.json

# the full catalog
curl https://d3codex.com/api/topics.json

# the LLM discovery file
curl https://d3codex.com/llms.txt

Reference content is the D3 10.3.4 manual merged with the 2016 edition; the text and marks belong to their respective owners. D3Codex is an independent, unofficial reference. See the disclaimer. Please be reasonable with automated requests.