Open source · Apache 2.0 · v0.1.0

The AI-native
programming language.

One canonical semantic source for people, IDEs, CI systems and autonomous agents—designed to produce deterministic programs with identities anyone can verify.

  • Deterministic
  • Capability-secure
  • Self-verifying
  • Content-addressed
examples/pipeline.axeph · excerpt AXEPH
; functions · bindings · explicit host effect
(module
  (def discounted
    (fn (subtotal)
      (if (> subtotal 1000)
        (- subtotal (/ subtotal 10))
        subtotal)))
  (def main
    (print (discounted 1788))))
$ axc run examples/pipeline.axeph
Axeph Foundry / total AUD 1610verified
full file identityax11e7df900fdd…cda10be

01 canonical semantics

02 machine-first truth

03 explicit effects

04 verifiable identity

Why Axeph?

Code that can explain
what it is.

Axeph begins with identity, effects and verification—the things conventional toolchains often bolt on after the language has already accumulated ambiguity.

01

Canonical by construction

Formatting, comments, definition order and bound-variable names converge on one normalised semantic representation.

implemented
02

Content is identity

Alpha-normalised AXB1 bytes are hashed with BLAKE3. Modules receive stable ax1… identities.

implemented
03

Effects have a door

The current print effect crosses an explicit Host boundary. Full inferred effect rows and runtime capabilities are next.

foundation shipped
04

Diagnostics are data

Stable codes, UTF-8 source spans and machine-applicable repairs make the checker useful to people and automated systems.

implemented
05

Hash-pinned artifacts

The local store writes, retrieves and re-verifies modules by semantic hash. Networked dependency resolution remains roadmap work.

local store shipped
06

One source, many targets

The interpreter is live today. AxIR, WASM, native and web backends will share the canonical semantic source.

roadmap
Architecture

One truth flows
through every tool.

The compiler pipeline keeps textual concerns at the edge. Canonical identity, checking and execution all work from the same AST.

Read the architecture notes
  1. 01
    Axeph sourcecompact s-expression syntax
  2. 02
    Spanned readerAST + exact UTF-8 locations
  3. 03
    Canonical corealpha-normalise · sort · AXB1
  4. 04
    Static checkerscope · arity · structured fixes
  5. 05
    Interpreter + Hostchecked evaluation · routed effects
  6. →
    Verification / targetsAxIR · WASM · native · web — planned
Genuine toolchain

Readable to humans.
Consumable by agents.

The CLI never maintains a second interpretation of the language. Canon, hash, check and run all call the same Rust core.

Semantic workflow5 commands

$ axc canon examples/pipeline.axeph

(module (def discounted …) (def receipt …) (def main …))

$ axc hash examples/pipeline.axeph

ax11e7df900fdd31…

$ axc check examples/pipeline.axeph

✓ well-formed

$ axc run examples/pipeline.axeph

Axeph Foundry / total AUD 1610

axeph-check · JSONmachine-first
{
  "severity": "error",
  "code": "AX0002",
  "message": "unbound variable 'prirnt'",
  "span": { "line": 4, "column": 21 },
  "fixes": [{
    "with_var": "print"
  }]
}
precise location stable code applicable fix
Built for humans and agents

One compiler truth.
Four kinds of consumer.

Developersclear CLI feedback
IDEsspans and repairs
CI systemsstable exit contracts
Agentsstructured verification

Agents do not scrape terminal prose. They receive the same structured evidence an IDE uses, then can apply a bounded repair and check again.

Axeph in KodeForge

A first-class editing
and verification loop.

Real Axeph language support: semantic highlighting, located diagnostics, execution, canonicalisation and content identity.

Axeph pipeline source with semantic syntax highlighting running successfully in KodeForge
01
Edit → check → runRevised semantic palette with genuine Axeph toolchain output
Axeph semantic highlighting and structured diagnostics displayed in KodeForge
02
Located diagnosticsStable codes, exact ranges and distinct semantic tokens
Axeph semantic syntax highlighting and content hash displayed in KodeForge
03
Semantic identityCanonical BLAKE3 identity on demand

Clear boundary: these screenshots demonstrate Axeph integration with KodeForge. KodeForge itself is proprietary and is not part of the open-source Axeph distribution.

Toolchain

Small pieces.
One semantic centre.

Seven focused components make the current language easy to inspect, test and evolve.

01axir

AST, reader, canonical writer, AXB1, normalisation, hash, checker and evaluator.

core
02axrt

Explicit runtime hosts for real and captured output.

runtime
03axc

Canonicalise, hash, binary, check and run from one driver.

cli
04axeph-play

Interactive REPL and fast file evaluation loop.

repl
05axeph-check

Machine-first JSON checker for IDE, CI and agent consumers.

verify
06axpm

Local semantic-hash object store with integrity verification.

store
07axeph-studio

Native dogfooding workbench built directly on axir.

ide
Open source

Build Axeph.
Inspect every layer.

The complete language and toolchain source is public under Apache License 2.0. KodeForge is not included.

Build from source
$ # Synchronise or download Axeph from KodeCryptX
$ cd Axeph
$ cargo build --workspace
$ cargo test --workspace --all-targets
$ ./target/debug/axc run examples/hello.axeph
Rust toolchain required Linux · macOS · Windows core
Current status

Early, working,
and honest about both.

Axeph 0.1 proves its semantic substrate through runnable artifacts. Roadmap claims do not graduate until their test gate is green.

01

Semantic substrate

Reader · canonical form · AXB1 · BLAKE3 · goldens

gate green
02

Pure core

Closures · recursion · checked primitives · REPL

gate green
03

Types & effects

Scope and arity shipped; inference and capability checks next

in progress
04+

Targets & verification

AxIR · WASM · native · contracts · package registry

roadmap
Read the complete test-gated roadmapAXEPH_CHECKLIST.md
AXEPH

Make the program
prove its identity.

Get AxephExplore source ↗
Copied to clipboard