Gortex — Code-Intelligence Engine for AI Agents
Gortex is a high-performance graph-based code-intelligence engine that cuts AI agent token usage up to 50× with 257-language support.
AI coding agents are powerful, but they waste tokens reading entire files when they only need a single function signature, a call chain, or an import graph. For large codebases, this inefficiency adds up fast.
Gortex solves this by indexing code into a persistent knowledge graph and exposing it through CLI, MCP Server, HTTP API, and a Web UI — giving AI agents surgical access to exactly the information they need.
What is Gortex?#
Gortex is a high-performance, graph-based code-intelligence engine built in Go. It indexes your codebase into a structured knowledge graph using tree-sitter AST analysis across 257 languages, enhanced with compiler-grade resolution for major languages. The result is a single static binary (zero dependencies) that serves as the brain for your AI coding assistant.
It supports 19 AI coding agents out of the box — Claude Code, Kiro, Cursor, Windsurf, VS Code / Copilot, Continue.dev, Cline, OpenCode, Antigravity, Codex CLI, Gemini CLI, Zed, Aider, and more. One gortex init configures every detected agent on your machine.
Why Gortex Matters#
Up to 50× Fewer Tokens#
Instead of feeding AI agents full source files, Gortex serves precise subgraph queries. The token savings dashboard shows real results:
Cost avoided: $168.69 (claude-opus-4) across 1,878 calls · 11,246,094 tokens saved
Today ████████░░░░░░░░ 50.0% saved 9,200 / 18,400 tokens
Last 7 days ██████████░░░░░░ 62.5% saved 60,100 / 96,200 tokens
All time ███████████████░ 93.3% saved 11,246,094 / 12,050,716 tokens $168.69257 Languages#
Three parsing tiers ensure every file in your repository is reachable:
- Bespoke tree-sitter grammars — deep AST analysis for major languages
- Regex-based extractors — broad coverage for smaller ecosystems
- Forest-backed signatures — fallback for anything not explicitly handled
Plus support for Jupyter and Databricks notebooks.
Multi-Repository by Default#
N repos in one graph — contracts, references, and call chains span repo boundaries with evidence-gated resolution, contract matching, and impact analysis.
Zero Dependencies#
Single static binary for macOS, Linux, and Windows. No Node.js, no Python, no Docker. No network required. Install and use instantly.
Key Features#
| Feature | Description |
|---|---|
| 257 language grammars | Tree-sitter AST + compiler-grade resolution for Python, TypeScript, Go, Rust, C/C++, Java, Kotlin, Swift, Zig, Ruby, and more |
| 175 MCP tools | Symbol lookup, call chains, blast radius, dataflow, clone detection, code actions — use only what you need |
| 19 agent integrations | One command configures every AI coding agent on your machine |
| Cross-repo API contracts | Auto-detects HTTP, gRPC, GraphQL, Kafka, WebSocket contracts across repos |
| Semantic search | Built-in GloVe-50d (3.8 MB, zero deps), opt-in MiniLM / Ollama / OpenAI |
| Blast radius analysis | Precomputed depth-3 reach index turns impact queries into O(seeds × reach) lookups |
| Speculative execution | preview_edit / simulate_chain answer “what would change?” without touching disk |
| Live editor overlays | Push unsaved buffers as a shadow graph; branching for parallel speculative sessions |
| GCX1 wire format | −27% tokens vs JSON at same fidelity |
| PR review | End-to-end: triage open PRs, emit line-anchored findings with BLOCK/REVIEW/APPROVE verdicts |
| Web UI | Standalone Next.js 15 UI with 3D force-directed graph visualization |
| Long-living daemon | One process serves every IDE window; live fsnotify, on-disk snapshots, restart |
Cross-Repo API Contracts#
Gortex auto-detects API contracts across repositories and matches providers to consumers:
| Contract type | Detection |
|---|---|
| HTTP routes | Framework annotations (gin, Express, FastAPI, Spring…) |
| gRPC | Proto service definitions |
| GraphQL | Schema type/field definitions |
| Message topics | Kafka / RabbitMQ / NATS / Redis pub/sub |
| WebSocket | Event emit/listen patterns |
| Env vars | os.Getenv, process.env, .env files |
| OpenAPI | Swagger / OpenAPI spec files |
| Temporal workflows | Go / Java SDK annotations |
Contracts are normalized to canonical IDs (e.g. http::GET::/api/users/{id}) and matched across repos to detect orphan providers/consumers and mismatches.
Performance Benchmarks#
Measured on an Apple Silicon laptop:
| Repository | Files | Nodes | Edges | Index time | Peak heap |
|---|---|---|---|---|---|
| torvalds/linux | 70,333 | 1,690,174 | 6,239,570 | ~3 min | 5.07 GB |
| microsoft/vscode | 10,762 | 204,501 | 808,902 | ~1 min | 580 MB |
| Gortex (self) | 430 | 5,583 | 53,830 | 3.4s | 52 MB |
Parsing dominates wall time (65–80%); reference resolution and search-index build scale sub-linearly.
Installation#
# macOS / Linux
curl -fsSL https://get.gortex.dev | sh
# Windows (PowerShell)
irm https://get.gortex.dev/install.ps1 | iexThe installer detects OS/arch, verifies SHA256 + cosign signatures, and installs to PATH.
Quick Start#
gortex install # One-time machine setup
gortex daemon start --detach # Start background daemon
gortex track ~/projects/myapp # Add a repository
cd ~/projects/myapp && gortex init # Per-repo configYour AI assistant now uses graph queries instead of raw file reads.
Architecture#
gortex binary
CLI (cobra) ──> MultiIndexer ──> In-Memory Graph (shared, per-repo indexed)
MCP (stdio) ──────────────────> Query Engine
HTTP /v1/* ────────────────────> Same tools + /v1/graph + /v1/events (SSE)
Daemon ────────────────────────> Shared graph for every MCP client
MultiWatcher <── filesystem events (fsnotify, per-repo)
CrossRepoResolver ──> Cross-repo edge creation
Persistence ──> gob+gzip snapshotGortex vs. Alternatives#
| Feature | Gortex | Graphify | Traditional RAG |
|---|---|---|---|
| Languages | 257 | 36 | N/A |
| Multi-repo | ✅ First-class | ❌ Single repo | ❌ |
| MCP tools | 175 | Limited | ❌ |
| Agent integrations | 19 | CLI skill | ❌ |
| Token savings | Up to 50× | Significant | ❌ |
| Cross-repo contracts | ✅ HTTP, gRPC, Kafka, etc. | ❌ | ❌ |
| Zero dependencies | ✅ Single Go binary | ✅ Python | ❌ Embeddings |
| Blast radius | ✅ O(seeds × reach) | ❌ | ❌ |
| Web UI | ✅ 3D force-directed graph | ✅ Interactive graph | ❌ |
| License | Apache 2.0 | Apache 2.0 | Varies |
Summary#
Gortex is a remarkably complete code-intelligence engine. With 257-language support, multi-repo by default, 175 MCP tools, 19 agent integrations, and proven token savings of up to 50×, it’s one of the most comprehensive tools available for making AI coding agents truly efficient on large codebases.
Its single-binary zero-dependency design means you can install and start saving tokens in minutes — no infrastructure, no dependencies, no cloud required.