Welcome to FoxNose Documentation

FoxNose is the serverless knowledge platform for building sophisticated AI applications and autonomous agents. It replaces the complex, fragmented stack of databases, search engines, ETL scripts, and hand-written agent tool servers with a single, powerful, and unified API.

The Challenge

Traditionally, building AI-powered applications and agents that rely on external knowledge involves stitching together multiple, disconnected services:

  1. Primary Data Store (e.g., Postgres): Holds your core business data.
  2. Full-Text Search Engine (e.g., Elasticsearch): For keyword-based search.
  3. Vector Database (e.g., Pinecone): For semantic, similarity-based search.
  4. ETL & Syncing Scripts: Complex logic to keep data synchronized across all three systems.
  5. Backend API: An orchestration layer that attempts to join results from different search indexes (keyword and vector) and enrich them with data from the primary store. This is often slow, complex, and delivers inconsistent results.
  6. Agent Integration Glue (custom MCP server, hand-written tool definitions): For each LLM or agent client you want to support, you build a tool-server layer that wraps the backend, defines tools manually, keeps them in sync with the schema as it evolves, and maintains a second access-control surface. Tool drift becomes a constant maintenance tax — and every schema change risks breaking your agents.

FoxNose: A Unified Solution

Instead of a scattered ecosystem, FoxNose provides a single, unified solution:

  • Unified Data & Search: Your data is stored once and is instantly available for filtering, keyword-based (full-text) search, pure semantic (vector) search, or a combination of both using our powerful hybrid search. No more data syncing or complex joins.
  • Built-in Embeddings: Automatically transform your content into vector embeddings on the fly.
  • Agent-Native API: Every Flux API exposes an embedded MCP server and route catalog out of the box. Connect Claude, an autonomous agent, or any MCP-compatible client directly to your content — no SDK wrapper, no glue layer.
  • Comprehensive Data Management: FoxNose provides essential features out-of-the-box, such as environments for different stages (development, staging, production), versioning of schemas and data for seamless evolution, robust data validation to ensure quality, and built-in localization for global reach. These are critical functionalities that would otherwise add significant complexity to a custom solution.
  • Serverless & Scalable: As a fully serverless platform, FoxNose handles all infrastructure concerns. You focus on building your AI application, not on managing servers, scaling databases, or worrying about uptime.
  • One Powerful API: The Flux API provides a single, consistent way to retrieve the exact information you need, perfectly ranked, without building a complex backend orchestration layer.

How It Works

FoxNose is built around a simple, powerful idea: a dual API architecture that separates the curation of knowledge from its retrieval and use in your applications.

1. Curate Your Knowledge with the Management API

The Management API is your control center for shaping and managing the "brain" of your AI. It allows you to:

  • Define Your Data: Create flexible content models using Folders and reusable Components.
  • Manage Content: Populate your knowledge base with Resources, with full support for versioning and schema migrations.
  • Ensure Quality: Use built-in data validation to maintain the integrity of your knowledge base.
  • Work Safely: Leverage isolated Environments (e.g., development, staging, production) for a professional and safe development workflow.
  • Go Global: Natively manage content in multiple languages with powerful Localization features.

This API provides the tools to meticulously curate the high-quality, structured knowledge that high-performance AI applications demand.

2. Deliver to Apps and Agents with the Flux API

Once your knowledge is curated, the Flux API delivers it — to applications, to AI agents, and to MCP-aware clients — through a single high-performance, read-only surface. The same prefix that serves REST traffic to your app serves MCP tools to your agent, with the same auth and permissions on both. It offers powerful, multi-faceted search:

  • Advanced Filtering: Build highly specific queries using structured filters with complex logical conditions.
  • Full-Text Search: Integrate classic keyword and phrase-based search.
  • Semantic Search: Go beyond keywords to find conceptually related content based on meaning and intent.
  • Hybrid Search: Combine all search methods in a single request to get perfectly ranked, relevant results.
  • Agent-Native Surface: Every Flux API ships with a per-prefix MCP server, router introspection, and schema introspection — letting AI agents discover your tool catalog at runtime without a custom integration. Hybrid search arrives at the agent as a single search tool.

Apps consume Flux as REST. Agents consume the same Flux as MCP tools. No second access-control layer to maintain, no parallel tool-server to deploy, no schema drift between them.

Where to Go Next?

Ready to get started? The best way to learn is by doing.

Was this page helpful?