ArangoDB

Prerequisites

• Search the current open issues

What are you trying to do that currently feels hard or impossible?

Describe the data source you’d like to see supported ArangoDB is a native multi‑model database that combines document, graph, and key/value data models in a single distributed engine, with a unified query language (AQL) and first‑class support for joins and traversals across different models. It is widely used in AI‑driven applications that need to mix graph reasoning, flexible document storage, and high‑performance search over heterogeneous data. ​

Why this is important for MCP Toolbox MCP Toolbox already shines when orchestrating multiple data sources and exposing them as consistent tools to LLMs and agents (AlloyDB/PostgreSQL, BigQuery, Elasticsearch, Neo4j, MongoDB, etc.). Adding ArangoDB as a first‑class source would unlock powerful scenarios where a single MCP server can coordinate: ​

multi‑hop graph traversals (knowledge graphs, entity‑relationship reasoning)

document‑centric RAG workflows over rich JSON structures

mixed queries that join graph and document data in one AQL query.

This fits very well with the positioning of MCP Toolbox as a unified, framework‑agnostic server for enterprise data in AI systems. ​

Existing MCP servers for ArangoDB There are already open source MCP servers for ArangoDB (for example ravenwits/mcp-server-arangodb and arango-mcp-server), which provide a solid proof of demand and show typical operations (CRUD, AQL queries, collection management, backup workflows, etc.). However, they are standalone servers, each with its own configuration and conventions, and do not benefit from the unified configuration, security model, and tooling ecosystem provided by MCP Toolbox. ​

How this could look in MCP Toolbox From a user perspective, an ArangoDB source could follow the same pattern as the existing database sources (kind: postgres, kind: neo4j, etc.), with something like kind: arangodb, pointing at an HTTP endpoint (single server or cluster) plus database credentials. Toolbox‑provided tools could then wrap common patterns such as: ​

execute AQL query with typed parameters

list/create/delete collections and graphs

run graph traversals with constraints

manage indexes and views relevant for AI/RAG use cases.

This would allow developers to plug ArangoDB into existing MCP Toolbox‑based agents with minimal extra code and reuse the same auth, auditing, and policy layers they already use for other databases. ​

Willingness to help I would be happy to help by testing an ArangoDB source and tools, providing real‑world AI/agent use cases, and validating configuration examples. If a design proposal or initial implementation draft is useful, I am also open to discussing it.

Suggested Solution(s)

No response

Alternatives Considered

No response

Additional Details

No response