Tools and processors reference

This page is the single entry point for everything Lucenia can do to your data: the tools an agent can call, the ingest processors that run while documents are written, and the search processors that run while queries are served.

note

The word tool means two different things in these docs. The Ingestion tools and utilities section covers third-party data shippers (Beats, Logstash, Fluentd, OpenTelemetry), the Lucenia CLI, and migration tooling — programs that run outside the cluster. This page covers the tools and processors that ship inside the cluster.

Ask the cluster, not the docs

A running cluster is always the authoritative inventory for the version you are on.

Surface	Request	Notes
Agent tools	GET /_plugins/_ml/tools	Returns name, type, description, and version for every registered tool type.
Ingest processors	GET /_nodes/ingest?filter_path=nodes.*.ingest.processors	Complete. Includes processors contributed by every installed module.
Search processors	GET /_nodes/search_pipelines	Incomplete — see the warning below.

warning

GET /_nodes/search_pipelines reports only request_processors and response_processors. It omits search phase results processors, which is where normalization-processor lives — the processor that makes hybrid search work. Do not treat its output as the full search processor list. The complete list is in Search processors.

Agent tools

A tool is a reusable building block an agent calls to perform one specific task — search an index, run a deployed model, call an external service. You reference a tool by its type when registering an agent. For parameters, examples, and per-tool detail, see Tools.

Tool type	What it does
AgentTool	Runs another agent by its agent ID.
CatIndexTool	Retrieves detailed index information for the Lucenia cluster (health, status, document counts, and store sizes).
ConnectorTool	Invokes an external service through a configured connector.
IndexMappingTool	Retrieves mapping and setting information for one or more indexes.
ListIndexTool	Lists the indexes in the cluster, along with their health, status, and document counts.
McpSseTool	Invokes a tool hosted on a remote Model Context Protocol (MCP) server. See Using MCP tools.
MLModelTool	Runs any deployed machine learning model by its model ID.
QueryPlanningTool	Turns a natural-language question into a Lucenia query (query DSL) using an LLM.
RAGTool	Retrieves with a k-NN search, then asks a generation model to answer using the retrieved context.
ReadFromScratchPadTool	Reads back the notes an agent saved to its per-conversation scratchpad.
SearchIndexTool	Searches an index using a query written in query domain-specific language (DSL).
VectorDBTool	Embeds your query text and runs a k-NN search, returning the matching documents as context.
VisualizationTool	Finds saved visualizations by matching a search term against their titles.
WriteToScratchPadTool	Saves a short note to an agent's per-conversation scratchpad for later recall.

note

Two names in the shipped ML Commons module do not match the type you write in a request:

• The class VisualizationsTool registers the type VisualizationTool (no s). Use the type.
• McpSseTool is registered as a type but is constructed by the cluster when an MCP connector discovers a remote tool. It is bound to an MCP client at that point, so it is not listed by the MCP built-in tool registry when unbound.

Ingest processors

Ingest processors transform documents on the write path, inside an ingest pipeline. The complete table, with a link to each processor's own page, is in Ingest processors.

The processors below have dedicated pages in this version:

Processor	What it does
chunk	Splits text content into overlapping chunks using recursive, fixed, semantic, or topic-shift algorithms. Designed for RAG and vector search pipelines.
content_extract	Extracts structured content blocks from documents in various formats (PDF, DOCX, HTML, images) via inline text, S3/HTTPS references, or base64 attachments.
embed	Generates vector embeddings from text, image, or multimodal content using Bedrock, OpenAI, or HTTP providers.
ellipse	Converts an ellipse geometry (WKT or GeoJSON) into a polygon approximation and indexes it as a geo_shape or shape. Useful for directional or asymmetric coverage areas such as cellular sectors.
image_tiling	Splits large images into fixed-size tiles for multimodal vector search. Supports GeoTIFF/COG with HTTP Range reads and geographic bounding box computation.
ocr	Performs optical character recognition on image blocks using LLM vision models (Claude on Bedrock). Handles charts, diagrams, tables, and handwriting.
rerank_prepare	Annotates chunks with document-level metadata and position scores for search-time reranking.

Search processors

Search processors transform requests, responses, and intermediate phase results on the read path, inside a search pipeline. The complete tables are in Search processors.

They run in three places:

• Request processors rewrite the query before it is executed — for example query_embedding, oversample.
• Response processors transform the hits that come back — for example multimodal_rerank, retrieval_grounding.
• Phase results processors run between search phases on the coordinating node. This is where normalization-processor lives, and it is the category GET /_nodes/search_pipelines does not report.

Related pages

• Agents and tools — agents and the tools they call.
• Ingest pipelines — building and running write-path pipelines.
• Search pipelines — building and running read-path pipelines.
• Ingestion tools and utilities — Beats, Logstash, the Lucenia CLI, and migration tooling.
• Nodes Info API — the API behind GET /_nodes/ingest.