Skip to content

Repository files navigation

OmniContext

A universal, natively-compiled semantic code context engine. Exposes structured codebase abstraction to AI agents through the Model Context Protocol (MCP).

OmniContext is engineered to perform high-speed code parsing, relationship extraction, and semantic embeddings locally, bridging repository structures with large language models seamlessly.

Status Version Build Platform License

Editor Extensions

You can install the official VS Code extension directly from the marketplaces:


Architecture Breakdown

Architecture Breakdown

  1. Local Execution: Embeddings (jina-embeddings-v2-base-code) and full database indexing execute locally. External cloud APIs are bypassed.
  2. Speed & Recall: Hybrid retrieval combines sub-millisecond AST boundary keyword matches with HNSW-optimized vector search and graph-boosted reranking.
  3. Language Support: Robust tree-sitter extraction across 16 languages (Rust, Python, TypeScript, Go, Java, C++, C#, CSS, etc.) and Markdown/TOML/JSON documentation.

Installation (Zero-Config)

Automated deployment logic resolves binaries natively from GitHub release artifacts. Execution auto-downloads the ONNX runtime library, pre-caches the embedding model (~550MB), and injects the MCP server manifest into active AI IDEs.

Windows (PowerShell):

irm https://raw.githubusercontent.com/steeltroops-ai/omnicontext/main/distribution/install.ps1 | iex

macOS / Linux (Bash):

curl -fsSL https://raw.githubusercontent.com/steeltroops-ai/omnicontext/main/distribution/install.sh | bash

Package Managers

Alternative lifecycle management via standard package managers:

# Windows (Scoop)
scoop bucket add omnicontext https://github.com/steeltroops-ai/omnicontext
scoop install omnicontext

# Windows (WinGet)
winget install omnicontext
# macOS (Homebrew)
brew tap steeltroops-ai/omnicontext
brew install omnicontext

# Cross-platform (Cargo / Pkgx)
cargo binstall omni-cli
pkgx install omnicontext

Uninstallation

To cleanly purge binaries and vector databases:

# Windows
irm https://raw.githubusercontent.com/steeltroops-ai/omnicontext/main/distribution/uninstall.ps1 | iex
# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/steeltroops-ai/omnicontext/main/distribution/uninstall.sh | bash

Core Usage

Re-initialize standard terminal sessions or source your shell RC (~/.zshrc or ~/.bashrc) before running commands.

# Verify components
omnicontext --version

# Index target repository (initializes SQLite and vector caches)
omnicontext index .

# Execute direct semantic search
omnicontext search "user authentication flow" --limit 5

# Launch headless daemon process (IPC)
omnicontext-daemon

MCP Integrations (Zero-Config)

OmniContext auto-discovers client configurations and writes its mcp-manifest.json path into their startup sequences upon install or daemon initialization.

Client Configuration Path Protocol Support
Claude Desktop claude_desktop_config.json Native stdio
Cursor cursor.mcp/config.json Native stdio
Windsurf mcp_config.json Native stdio
Cline / RooCode mcp_settings.json Native stdio
Trae IDE .trae/mcp.json Native stdio
Kiro mcp.json Native stdio (powers.mcpServers)
Continue.dev config.json Native stdio
Antigravity mcp_config.json Native stdio
Claude Code (CLI) ~/.claude.json Native stdio

Development & Contribution

OmniContext strictly enforces structural standards, code formatting, and conventional commits. Bypassing formatting or naming standards will block CI chains.

Local Building

# Clone source mechanism
git clone https://github.com/steeltroops-ai/omnicontext.git
cd omnicontext

# Build full workspace
cargo build --workspace --release

# Execute CI validation matrices locally
cargo fmt --all -- --check
cargo clippy --workspace -- -D warnings
cargo test --workspace

Commit Protocol

Commit messages must conform strictly to the Conventional Commits specification. The git-cliff auto-changelog processor relies entirely on these prefixes:

  • feat(scope): - New features (major version bump if marked with !)
  • fix(scope): - Bug corrections
  • perf(scope): - Optimizations
  • refactor(scope): - Structural changes without feature addition
  • docs(scope): - README or documentation edits
  • chore(release): - Internal/CI modifications (excluded from changelog)

For detailed formatting constraints and expected boundaries, review .agents/workflows/commit-conventions.md.


License

OmniContext tools and base engine components are provided under the terms of the Apache License version 2.0. See LICENSE for full details.

About

A universal, natively-compiled semantic code context engine. Exposes structured codebase abstraction to AI agents through the Model Context Protocol (MCP).

Resources

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages