Stop AI Hallucinations for Good
Yavy turns your documentation into a semantic knowledge base that AI assistants can use. No more made-up answers - just real information from your real content.
Core Features
Make your documentation AI-searchable
Developer CLI
The fastest way to use Yavy. Install the CLI to search docs, list projects, and generate Skills - all from the terminal. Perfect for CI/CD.
- npm install -g @yavydev/cli
- Search & generate
- CI/CD friendly
Portable Skills
Generate skill packages with the Yavy CLI or Claude Code plugin. Drop them into any project for offline AI access - no server needed at runtime.
- CLI & plugin generation
- Offline capable
- Version controllable
Semantic Vector Search
Our search understands meaning, not just keywords. Find relevant content even when queries use different terminology than your documentation.
- Vector embeddings
- Contextual understanding
- Multi-language support
Automatic URL Discovery
Just add your website URL - we automatically discover all pages via sitemaps, robots.txt, and recursive crawling. No manual page-by-page entry required.
- Sitemap.xml parsing
- Recursive crawling
- robots.txt respect
Instant AI Integration
Every project gets CLI access, Skills export, and a unique MCP server endpoint. Connect any AI assistant in minutes using your preferred method.
- CLI search
- Skills package export
- MCP server protocol
Real-Time Content Sync
Your AI assistant always has the latest information. We automatically detect content changes and update embeddings accordingly.
- Automatic re-indexing
- Change detection
- Manual sync option
Multi-Source Indexing
Index content from multiple sources: websites, GitHub repositories, Notion workspaces, and Confluence spaces. All in one unified search.
- GitHub integration
- Notion workspaces
- Confluence spaces
Private Content Support
Connect OAuth providers to index private documentation. Your AI assistant can search internal wikis and protected content securely.
- OAuth integration
- HTTP auth support
- Custom headers
How It Works
From content source to AI-searchable knowledge in minutes
Add Your Source
Add a website URL, connect a GitHub repository, Notion workspace, or Confluence space. We automatically discover and index all content.
We Index Everything
Content is crawled, processed, and converted into semantic vector embeddings. This enables meaning-based search that understands context.
Connect Your AI
Install the CLI and start searching immediately. Generate a Skills package for offline access, or connect via MCP for real-time search. Your choice.
Works With Your Tools
Connect via CLI, Skills, or MCP to any AI assistant
Claude Desktop
Anthropic's official desktop app
Claude Code
MCP + Skills support
Cursor
AI-first code editor
Windsurf
AI coding companion
VS Code
With MCP extensions
ChatGPT
OpenAI's AI assistant
Codex CLI
OpenAI's coding agent
OpenCode
Open-source terminal AI
Custom Clients
Any CLI, Skills, or MCP-compatible tool
Built for Developers
Technical specifications and capabilities
Content Processing
- HTML content extraction with semantic structure preservation
- Markdown, code blocks, and technical documentation support
- Automatic chunking optimized for LLM context windows
- Metadata extraction (titles, headings, timestamps)
Search Capabilities
- High-dimensional vector embeddings for semantic similarity
- Hybrid search combining vector and keyword matching
- Configurable relevance thresholds and result limits
- Source attribution with page URLs and titles
Infrastructure
- Global edge network for low-latency MCP responses
- Automatic scaling based on query volume
- High availability infrastructure
- Enterprise-grade data handling
Security
- OAuth 2.1 authentication for MCP connections
- Project-level access controls and team permissions
- Encrypted data at rest and in transit
- Granular access controls