Gortex

// Overview

Gortex is a code-intelligence engine built for AI coding agents and IDEs. It parses a codebase with tree-sitter (the README counts 257 languages and grammars), adds compiler-grade resolution for languages such as Python, TypeScript, Go, Java, Rust and C#, and stores the result as a persistent knowledge graph of functions, classes, call chains, HTTP routes and cross-service contracts. Instead of reading whole files, an agent asks the graph for exactly the symbol, caller or blast radius it needs — the project's pitch is up to 50x fewer tokens per response, with the benchmarks published in the repository.

It ships as a single static Go binary for macOS, Linux and Windows with no external dependencies. A long-running daemon serves one shared graph to every editor window, watches files for changes and keeps the graph in an on-disk SQLite store. The same graph is reachable three ways: the `gortex` CLI, an MCP server, and a versioned `/v1/*` HTTP API that also backs a separate Next.js web UI for exploring the graph. Multi-repository workspaces are the default, so references and API contracts can be followed across repo boundaries.

On top of the graph sit higher-level features: default-on semantic search with an embedded GloVe model (MiniLM, Ollama or OpenAI are opt-in), blast-radius and dataflow queries, speculative `preview_edit` checks, PR triage and review commands, and a `gortex savings` report of tokens saved versus naive file reads. `gortex init` configures the coding agents it detects on the machine — the README lists 19, including Claude Code, Cursor, Windsurf, VS Code / Copilot, Codex CLI, Gemini CLI, Zed and Aider. Telemetry is off by default, and Gortex is released under the Apache 2.0 license.

// What it does

• Indexes code into a persistent knowledge graph of functions, classes, call chains, HTTP routes and cross-service contracts, parsed with tree-sitter across 257 languages and grammars
• Multi-repo by default: one graph spans many repositories, with cross-repo references, impact analysis and API-contract matching for HTTP, gRPC, GraphQL, message topics, WebSocket, env vars, OpenAPI and Temporal
• Serves the graph over an MCP server, the CLI and a versioned HTTP API with MCP Streamable HTTP transport, plus a standalone web UI with 3D graph views
• Semantic search on by default using an embedded GloVe model with FTS5/BM25 plus vector fusion, and optional MiniLM, Ollama or OpenAI embeddings
• PR workflow commands: `gortex prs` triages open pull requests by blast radius and `gortex review` emits line-anchored findings with a BLOCK / REVIEW / APPROVE verdict
• Single static binary with a background daemon, live file watching and an on-disk SQLite graph store; `gortex savings` reports tokens saved versus reading files

// Getting started

Gortex installs as one signed binary. The README's quick start is four commands: set up the machine once, start the daemon, track a repo, and initialise that repo for your agents.

# macOS / Linux
curl -fsSL https://get.gortex.dev | sh

# Windows (PowerShell)
irm https://get.gortex.dev/install.ps1 | iex

gortex install

gortex daemon start --detach
gortex track ~/projects/myapp

cd ~/projects/myapp && gortex init

gortex daemon start --http-addr 127.0.0.1:7411

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Cut the tokens a coding agent spends on large codebases by answering symbol, caller and blast-radius questions from a graph instead of whole-file reads
• Trace API contracts and call chains across several microservice repositories at once, and spot orphaned providers or consumers
• Check what an edit would break before applying it, with blast-radius, `preview_edit` and `verify_change` queries
• Triage and review open pull requests with graph-grounded risk scores and line-anchored findings

// How Gortex compares

Gortex alongside other open-source code indexing & search tools AI/TLDR tracks, ranked by GitHub stars.