Compile
Strong static types, algebraic effects, pattern matching. Two backends: QBE for fast iteration, LLVM for production binaries.
(
)
=>
Axon Programming Language
A compiled language for the AI-native era.
Strong types, zero-cost actors, end-to-end verification, and a package manager that ships binaries — built so humans and agents can ship the same production code.
Verify
Linear types for resource ownership, runtime contracts, deterministic replay debugger. Catch correctness bugs at compile time, runtime issues at record time.
Distribute
axpm package
manager with content-addressed registries, deterministic lockfiles,
and Ed25519 package signing for supply-chain security.
benchmark
—
0-commit trend
typecheck
0
modules checked
coverage
0%
lines · fn 0%
tests
—
no junit reported
What's new
- 1mo agofeat(analytics): charter, dashboard, and metrics fetcher (#1)axon-analytics →
- 1mo agofeat(dap): minimal Debug Adapter Protocol server for Axon (#123)axon-debug →
- 1mo agofeat(scaffold): add `axonc new`/`axonc init` project scaffolder (#490)axon-lang →
- 1mo agofeat(memprof): expand to full memory profiler + leak detector (#121)axon-debug →
- 1mo agofeat(outdated): add `axpm outdated` semver-aware upgrade checker (#106)axon-pkg →
AXON
The AI-Native Systems Language.
Built for machines to write, humans to scale.
Re-engineering the Foundation
Axon discards the complexity of legacy systems languages, offering a minimalist, deterministic foundation optimized entirely for machine generation and hyper-scale concurrency.
01. The Token Crisis
Large Language Models generate code by predicting the next token. When they generate C++ or Rust, they must navigate a minefield of complex, human-ergonomic syntax operators. Misplaced brackets, borrowing rules, and operator precedence cause hallucinations and build breakages.
02. Homoiconic Elegance
Axon uses S-expressions. The code structure is exactly identical to the Abstract Syntax Tree. There is zero ambiguity. It requires roughly 40% fewer LLM tokens to generate logic in Axon than in Rust, significantly expanding the effective context window.
03. Uncompromising Speed
Despite its LISP-like syntax, Axon is not interpreted. It compiles directly to native machine code via the LLVM backend. It features absolute zero Garbage Collection, utilizing an M:N Actor model for lightweight concurrency that scales effortlessly across CPU cores.
See the Syntax
Homoiconic S-expressions make parsing trivial for AI models, guaranteeing zero ambiguity and eliminating entire classes of token generation errors.