API Reference

Supported Languages

OmniContext provides full AST parsing support for 13+ programming languages using tree-sitter grammars. Each language includes symbol extraction, import resolution for dependency graphs, visibility inference, and documentation extraction.

Language Support Matrix

Language	tree-sitter Grammar	AST Extractor	Graph Resolver	Tests	Status
Python	`tree-sitter-python`	✓	✓	✓	Core Baseline
TypeScript	`tree-sitter-typescript`	✓	✓	✓	Core Baseline
JavaScript	`tree-sitter-javascript`	✓	✓	✓	Core Baseline
Rust	`tree-sitter-rust`	✓	✓	✓	Core Baseline
Go	`tree-sitter-go`	✓	✓	✓	Core Baseline
Java	`tree-sitter-java`	✓	✓	✓	Extended
C	`tree-sitter-c`	✓	✓	✓	Extended
C++	`tree-sitter-cpp`	✓	✓	✓	Extended
C#	`tree-sitter-c-sharp`	✓	✓	✓	Extended
CSS	`tree-sitter-css`	✓	—	✓	Extended
Ruby	`tree-sitter-ruby`	✓	✓	✓	Pending Baseline
PHP	`tree-sitter-php`	✓	✓	✓	Pending Baseline
Swift	`tree-sitter-swift`	✓	✓	✓	Pending Baseline
Kotlin	`tree-sitter-kotlin-ng`	✓	✓	✓	Pending Baseline

Status meanings:

Core Baseline — full support, thoroughly tested, production-ready.

Extended — full support, tested, actively maintained.

Pending Baseline — functional, test suite being expanded.

Language Capabilities

1. AST Structural Mapping

Each language parser maps tree-sitter nodes to OmniContext's internal chunk representation. The following symbol kinds are extracted:

Kind	Description
`function`	Function declarations and definitions
`class`	Class declarations
`trait`	Trait and interface definitions
`impl`	Implementation blocks (Rust)
`const`	Constants and static values
`type`	Type aliases and definitions
`module`	Module and namespace declarations
`test`	Test functions and test blocks

2. Graph Import Resolution

OmniContext builds a dependency graph by resolving language-specific import statements at index time:

Language	Resolved Constructs
Python	`import`, `from ... import`, `importlib`
TypeScript / JavaScript	`import`, `require()`, barrel re-exports (`export * from`)
Rust	`use`, `mod`, `pub use`
Go	`import`
Java	`import`, `package`
C / C++	`#include`
C#	`using`

3. Visibility Inference

Search results and dependency graphs respect language-specific visibility rules:

Language	Visibility Mechanism
Python	Heuristic: names without a leading `_` are treated as public
TypeScript / JavaScript	Static: presence of `export` token
Rust	Explicit: `pub`, `pub(crate)`, `pub(super)`
Go	Syntactic: capitalized identifiers are public
Java	Keywords: `public`, `private`, `protected`, package-private
C#	Keywords: `public`, `private`, `internal`, `protected`

4. Documentation Extraction

OmniContext extracts documentation comments and makes them searchable alongside symbol names:

Language	Comment Format
Python	`"""docstring"""`
TypeScript / JavaScript	`/** JSDoc */`
Rust	`/// doc comment` and `//! inner doc comment`
Go	`// Package / function comment` (preceding the declaration)
Java	`/** Javadoc */`
C#	`/// <summary>XML doc comment</summary>`

Adding New Languages

To add support for a new language, follow the workflow defined in .agents/workflows/add-language.md. The average integration time is approximately 3 engineering days.

Required steps:

Add the tree-sitter grammar crate to Cargo.toml in crates/omni-core.
Implement the LanguageParser trait in crates/omni-core/src/parser/languages/.
Add graph import resolution logic.
Register the language in crates/omni-core/src/parser/registry.rs.
Create unit test fixtures in crates/omni-core/tests/.
Submit a pull request with the new language and test fixtures.

Performance

All language parsers are designed to maintain consistent throughput:

Metric	Target
File parsing	> 500 files / second
AST extraction	< 5 ms per file
Graph resolution	< 10 ms per file
Chunk memory overhead	< 2 KB per indexed chunk

These figures are measured on modern commodity hardware (8-core CPU, NVMe storage) using the core baseline languages.