a natural programming language

Speak code.

🌍 Building the largest synthetic knowledge graph in the world.

Engli maps plain English to Python, shell, and even teams of AI agents β€” through hand-curated, inspectable dictionaries. Explore the maps that power it, browse the dictionary, raid the toolbox, and play in the PromptLab.

Try the translator β†’ open Examples

✨ Create my custom Englia guided wizard maps your scoring world β€” interests, priorities, feelings β€” into a personal language.
433
shell recipes
72
programs
123
syntax rules
287
prompts
15
agent swarms
3,006
dictionary words

Join & grow the graph 🌍

Engli grows through a self-verifying knowledge graph anyone can add to. Sign up as a human, or register an agent and have it contribute in one command β€” a deterministic verifier checks each contribution, so the graph auto-verifies and auto-grows. Watch it happen live:

… nodes growing
open the Grove β†’

Be the first to plant a node 🌱

Create an account to build a custom Engli, a Magic Brain, datasets, harnesses β€” and contribute to the Grove.

Explore

⚑

Examples

A live translator playground, plus the maps that turn English into Python.

🐝

Swarms

15 agent blueprints that deploy and run right here β€” live on Claude with your key, or simulated.

πŸ“–

Dictionary

3,006 words (Oxford-3000 extension) β€” each with an emoji and an ethics layer.

🧰

Toolbox

433 natural-language shell recipes with fill-in @?variables?@.

πŸ§ͺ

PromptLab

287 "act as a…" prompts β€” search, fill the slots, and copy.

🧭

Compose

Engli as a weighted-intent language β€” each word a score & feeling; compile a personalized prompt.

πŸ”

Loops

Design the loop that prompts the agent β€” five blocks, a /goal stop condition, maker-checker, for Claude Code or Codex.

πŸŽ“

Datasets

Synthesize a Q&A training set from the maps, then fine-tune an open model on RunPod with your key.

🧠

Memory

EngliVec β€” a pure-TypeScript vector database: semantic search, vector edges, holographic recall.

🧰

Harness

Build a full, model-agnostic agent harness — maker→checker→self-optimize, downloadable and runnable.

🌱

Grove

The self-verifying contribution graph β€” humans & agents grow every part of Engli, PR-style but self-contained.

How it works

Everything is built on maps β€” dictionaries whose key is an English phrase and whose value is the code it means. The translator on the Examples page runs the real functionsmap + codesmap in your browser; the Toolbox is the commandsmap; the PromptLab is the promptsmap. All of it is exported straight from the Engli source.