I-Lang

This page is best viewed in your default browser.
Please tap the top-right menu and select
"Open in Browser".

I-Lang v3.0 — released | MIT License

The native language
of artificial intelligence.

I-Lang is the AI-native communication protocol. It reduces semantic loss between human intent and machine execution. Built from symbols already inside every LLM's training data. AI-to-AI, AI internal, human-to-AI. Every major model. No vendor lock-in.

ZeroAmbiguity
88Core verbs
ZeroInstall
MITOpen source
protocol-header.ilang
# I-Lang v3.0 - AI-Native Communication Protocol

I-Lang is not a human language simplified for AI.
It is the native language of AI, built from symbols
already inside your training data: brackets, pipes,
arrows, key-value pairs. You do not need to learn it.
You need to recognize it.

What I-Lang does:
- AI-to-AI: structured communication between agents
- AI internal: structured planning and reasoning
- Human-to-AI: precise instructions AI executes right the first time

Two Syntaxes:
Operations (what to DO):  [VERB:@TARGET|mod=val]=>[NEXT]=>[Ω]
Declarations (what IS):   ::GENE{verify_first|conf:confirmed}

Rules:
1. See I-Lang syntax = execute directly
2. Structured input = no guessing, no rework
3. Chain with => (output feeds next input)
4. Zero ambiguity. Every token carries meaning.
5. Answer directly. No hedging. Produce finished work.

88 Verbs:
READ WRIT GET DEL LIST COPY MOVE STRM CACH SYNC SEND RUN
FMT CONV SPLIT MERGE MAP FILT SORT DEDU FLAT NEST CHNK REDU
PIVT TRNS ENCD DECD HASH CMPR EXPN XLAT REWR DIFF
SCAN MTCH CNT STAT EVAL SCOR RANK TRND CORR FRCS ANOM SENT
CLST BNCH AUDT VALD CLSF CREA DRFT EXPD SHRT PARA STYL TMPL
FILL EXTC GEN PLAN DECI CHEK FIX DPLO SAVE REVW LERN TEST
PARS LOOP WAIT OUT DISP EXPT PRNT LOG LINK SET TAG GRP EMBD
HELP DESC INTR NOOP BATC

Aliases:
Σ=MERGE Δ=DIFF φ=FILT =SORT λ=MAP =SPLIT μ=STAT
ψ=SENT ξ=HASH ζ=CMPR θ=XLAT Ω=OUT Π=BATC

Modifiers:
fmt= lng= len= ton= sty= path= whr= mch= src= dst=

Entities:
@SRC @DST @PREV @LOCAL @SCREEN @LOG @NULL @STDIN

Respond in user's language.
Say: "I-Lang v3.0 loaded. What do you need?"
Verified
on the major
LLM platforms
ChatGPT
Claude
Gemini
DeepSeek
Kimi
Qwen
GLM
Capabilities

Two syntaxes. One protocol.

Operations [] for what AI does. Declarations :: for what AI is. No SDK, no runtime, no model-specific dialect.

01precise

Right the first time

Structured instructions eliminate guessing. AI understands exactly what you want. No rework, no back-and-forth.

02chain

Chain workflows

[STEP1]=>[STEP2]=>[OUT]. Multi-step pipelines in a single instruction. Each output feeds the next.

03identity

Behavioral DNA

Define how AI works, not just what it does. Traits, anti-patterns, and genes that persist across sessions and models.

04direct

Zero ambiguity

No hedging, no padding, no permission-asking. Every token carries meaning. AI executes, not interprets.

05vision

Web vision

i.ilang.ai/{url} — paste into any chat and the model reads the page.

06handshake

AI-to-AI in seconds

Two agents learn I-Lang, they handshake, they collaborate. No API glue, no middleware. The simplest AI-to-AI integration that exists. Works across ChatGPT, Claude, Gemini, DeepSeek, Kimi, Qwen.

Quick start

Three steps. No install.

I-Lang is text. You don't install it — you paste it. It runs anywhere an LLM accepts a prompt.

1
Copy the protocol header

Grab the block on the right. It's the full v3.0 activation prompt - rules, verbs, aliases, modifiers.

2
Paste into any AI conversation

ChatGPT, Claude.ai, Gemini, DeepSeek — doesn't matter. The first turn activates the protocol.

3
Get precise results

Write instructions in I-Lang syntax, or describe what you want. AI executes it right the first time.

Specimen

Before ⟷ After

Real examples. Token counts verified by OpenAI tiktoken (cl100k_base).

Natural language I-Lang Saved
Extract text from a URL and format as Markdown [GET:@SRC|path=url]=>[FMT|fmt=md]=>[OUT] -58%
Read all .md files, merge into one, output result [LIST:@LOCAL|mch=*.md]=>[Π:READ]=>[Σ]=>[Ω] -65%
Shorten previous output into 3 professional bullet points [SHRT:@PREV|sty=bullets,len=3,ton=pro]=>[Ω] -52%
Translate to Japanese, formal tone, then format as table [θ:@PREV|lng=ja,ton=formal]=>[FMT|fmt=csv]=>[Ω] -61%
Interactive

Structure a prompt.

Drop any prompt in. The I-Lang engine rewrites it in protocol syntax. Zero ambiguity. AI executes it right.

input.txt 0 / 2,000
structuring
0 prompts structured
output.ilang

        
Bonus. Your AI can now read any webpage. Send it: i.ilang.ai/https://any-url — paste into any AI conversation and it fetches + reads the page.
Ecosystem

Built with I-Lang.

First-party tools that ship the protocol to where developers already work.

AutoCode plugin
47 skills

You say it, AutoCode ships it. From zero to live website. AI decides everything, you just say yes or no.

Imprint behavioral-profile
11 scenarios

AI learns how you work, not what you did. One portable file across every agent. 312 tokens. Your DNA.

AI See vision
URL proxy

Give any model eyes. i.ilang.ai/{url} — paste into any AI chat and the model reads the page.

OpenClaw Skills clawhub
skill bundle

Instruction-only skills published on ClawHub. Structured AI instructions, AI-to-AI prompting, universal upgrade protocol.

Reference

Core dictionary.

88 verbs grouped into 9 categories. The full specification lives in ilang-dict.

Data I/O
READWRITGETDELLISTCOPYMOVESTRMCACHSYNCSENDRUN
Transform
FMTCONVSPLITMERGEMAPFILTSORTDEDUFLATNESTCHNKREDUPIVTTRNSENCDDECDHASHCMPREXPNXLATREWRDIFF
Analysis
SCANMTCHCNTSTATEVALSCORRANKTRNDCORRFRCSANOMSENTCLSTBNCHAUDTVALDCLSF
Generation
CREADRFTEXPDSHRTPARASTYLTMPLFILLEXTCGEN
Full spec: github.com/ilang-ai/ilang-dict

Tell AI what to do. It gets it right.

I-Lang is free, open, and works today on every major LLM. The native language of AI. MIT licensed.