Documentation hub for an AI-native programming language in research.
This hub links what exists now and labels what is still planned. The official GitHub repository is available, while install, CLI, parser, and standard-library claims remain pending until implementation evidence exists.
Build the native runtime, create a .fx file, run it and inspect it with felidae_debug.
Implementation-oriented reference for current facts, methods, queries, imports and runtime behavior.
SentencePiece integer tokenization, IntegerParser, AST, interpreter and native-module boundaries.
Structured diagnostics, symbol/operator metadata and JSON-RPC editor integration.
Current newline-terminated Felidae programs used as the grammar/reference corpus.
Native declarations, loading boundary and C++ module contract.
Visualization/analytics implementation and self-contained HTML generation.
VS Code, IntelliJ and additional editor integration sources.
Implementation-first status with research direction separated from current runtime claims.
Active Felidae source repository, examples, tests, tools and editor integrations.
Official Source
The active repository is github.com/xnvtserver/Felidae.
Getting Started
Build the native CMake target, run a current v2 .fx example, execute a query and inspect it with felidae_debug.
Installation / Build
Native Linux/macOS/Windows build wrappers exist. Stable release distribution and non-native toolchains remain active engineering work.
Language Essentials
Current syntax is represented by v2_examples and the IntegerParser implementation, including facts, methods, queries, arrays, lambdas, inheritance and custom operators.
Fact / Query Runtime
Loaded facts participate in explicit query solving and indexed runtime access rather than requiring a separate SQL translation layer.
Native Modules
Typed .fx contracts connect to C++ shared-library targets for host-facing and specialized capabilities.
Debugging and LSP
felidae_debug owns AST diagnostics, --check-json, --lsp and editor metadata endpoints.
Celidae
Nine structural/analytical view implementations and acceptance tests exist; current CMake build-target integration is incomplete.
Editor Tooling
VS Code and IntelliJ have richer integrations; additional lightweight editor modes and Tree-sitter work are present at different maturity levels.
Platforms
Native host builds are the supported wrapper path today; non-native targets need proper CMake toolchains/presets.
AI-native Research Direction
Agent abstractions, richer memory/tool/policy semantics and higher-level AI orchestration remain explicitly separated from current compiler-supported syntax.