Skip to content

Semantic catalogue and operation discovery

The semantic catalogue is the machine-readable description of streamfind's public interface. It defines operation names, domains, parameters, typed input and output ports, constraints, result fields, units, nullability, and usage guidance.

What the catalogue provides

The catalogue allows an application or AI agent to discover:

  • which Operations can be called directly;
  • which operations are available;
  • required and optional parameters;
  • nested object and array schemas;
  • defaults, examples, constraints, and units;
  • result shapes and project effects;
  • suggested next operations and whether a connection is required.

The browser application, C++ API, and MCP server use the catalogue to present tool names, descriptions, input schemas, ports, and result contracts.

Operations and operation graphs

  • Operations are callable project or domain actions. They expose typed ports and operation-specific parameters.
  • Operation graphs persist operation nodes and typed connections. A node stores its operation identifier and JSON parameters.

Typical project entry points are:

create -> describe -> get_domain/get_metadata
      -> get_operations -> get_operation
      -> add_operation -> connect_operations
      -> validate_workflow -> run_workflow
      -> get_result_registry
        -> close

Runtime catalogue

The native packages include the catalogue under core/. The MCP server needs catalogue.duckdb at runtime. An explicit STREAMFIND_CATALOGUE path can be used when an application manages the catalogue separately.

See Releases for the package contents and How streamfind works for the relationship between the catalogue, native backends, and MCP.