Open Index
Build a searchable, continuously updated structured domain context layer for AI agents
About Open Index
Open Index is a structured context layer for AI agents, used to build domain-specific, searchable, and continuously improved context graphs. You can define concept relationships in support organizations, sales processes, or infrastructure, while the system stores, searches, and visualizes the relationship map. It provides read-write context to domain agents through MCP, and `--read-only` can restrict it to read-only; local storage uses SQLite + FTS5 by default, with support for an OpenSearch backend as well.
Best for
Best use cases
Build customer support knowledge graphs
Manage sales process relationships
Connect infrastructure services with runbooks
Provide domain context for legal or marketing agents
Features
Context graph
Organizes domain concepts and relationships into a searchable, continuously improved context graph.
MCP read-write
Provides structured context to domain-specific agents through MCP, with read-write enabled by default and a read-only mode available.
Local storage
Uses SQLite + FTS5 by default and runs with zero external services.
Pluggable backends
Backend interfaces support both SQLite and OpenSearch implementations.
Pros
- Uses SQLite + FTS5 by default, with no external services required
- Context graph is searchable and continuously improved
- Supports both SQLite and OpenSearch backends
- Use `--read-only` to prevent agents from modifying the context
Cons
- SQLite backend performs brute-force cosine scans over entities, and is suitable for about 10,000 entities or fewer
- Without a token configured, the HTTP service is unauthenticated, so users who can access the port can write to the context
Pricing Plans
MIT License
- Freely obtain, use, copy, modify, publish, sublicense, and sell
* Pricing information is for reference only. Please refer to the latest announcement on Open Index's official website.