Tocharian builds and maintains open-source MCP servers that let AI agents work safely across Elastic, security, cloud, and data platforms. Our work centers on three directions:
Access, search, and manage Kibana from any MCP-compatible client. The most widely used project we maintain.
Query and manage Elasticsearch from MCP-capable AI clients, with a tool surface built around security analysis and incident investigation.
An evaluation harness that measures whether an AI agent can actually run a security investigation in Elasticsearch — labeled attack telemetry, a graded exam, and a score.
A growing set of open-source MCP servers, each in its own repository. Every link goes straight to the source.
A fuzzy-search plugin for Elasticsearch tuned for typo-tolerant retrieval, such as license-plate and store-name search in e-commerce.
Generates diagnostic reports for Elasticsearch clusters to support health checks and performance reviews.
Also available as a free hosted tool ↗Collects Mac temperature, fan speed, and CPU thermal metrics into Elasticsearch/Kibana dashboards for analysis.
Elasticsearch in enterprise environments is not always on the latest version. Tocharian’s open-source and commercial engineering work can cover:
Enterprises do not need to rebuild their entire Elastic environment just to use AI or MCP.
Enterprises can get commercial support built on Tocharian’s open-source projects. Services include:
MCP lets AI call enterprise tools, but tool-calling must be tightly controlled. In enterprise MCP projects, Tocharian prioritizes:
AI can assist with analysis and prepare actions, but it must not bypass the customer’s existing security controls.
Open-source projects provide general capability. Enterprise projects usually also need to solve:
Tocharian helps enterprises turn open-source tools into complete, production-ready implementations.