context7.yaml

# Context7 MCP baseline configuration for the hgraph project
# This config helps the indexer prioritize core source files, reduce noise,
# and generate better summaries and cross-references.

project:
  name: hgraph
  description: >-
    HGraph is a functional reactive programming engine and DSL in Python for building
    time-series based computation graphs supporting simulation and real-time event processing.
  homepage: https://hgraph.readthedocs.io/en/latest/

indexing:
  include:
    - README.rst
    - pyproject.toml
    - src/hgraph/**
    - docs/**
  exclude:
    - ".git/**"
    - ".idea/**"
    - ".vscode/**"
    - ".venv/**"
    - "venv/**"
    - ".mypy_cache/**"
    - ".pytest_cache/**"
    - "dist/**"
    - "build/**"
    - "site/**"
    - "docs/_build/**"
    - "**/__pycache__/**"
    - "**/*.pyc"
    - "uv.lock"
    # De-prioritize by exclusion; examples and tests can be included with lower priority if desired
    - "docs_md/**"
  # Files that are allowed but should be lower priority for retrieval
  low_priority:
    - examples/**
    - hgraph_unit_tests/**

priorities:
  # Higher numbers mean higher priority
  - pattern: src/hgraph/**
    score: 1.0
  - pattern: docs/**
    score: 0.8
  - pattern: 'README.*'
    score: 0.7
  - pattern: examples/**
    score: 0.4
  - pattern: hgraph_unit_tests/**
    score: 0.4

chunking:
  code:
    max_lines: 200
    overlap_lines: 20
  markdown:
    max_chars: 2000
    overlap_chars: 200

summarization:
  enabled: true
  per_file: true
  hints:
    - Prefer summarizing public APIs, decorators, and type aliases exposed by hgraph.
    - Capture the meaning of TS, TSL, TSB, TSD, TSS, REF, SCALAR, OUT, TSW, and TIME_SERIES_TYPE type constructs.
    - Capture the meaning of compute_node, sink_node, graph, generator, and push_queue decorators.
    - Note relationships between nodes, graphs, schedulers, and the runtime.
    - For operator modules, summarize available operators and their signatures, note that operators are implemented when 
      the overloads attribute is defined in a graph or node decorator. Overloads are not called directly, but the 
      differences in signature and behaviour is useful to aggregate under the operator function.

cross_reference:
  symbols: true
  link_code_to_docs: true

entrypoints:
  - src/hgraph/__init__.py
  - README.rst

terminology:
  glossary:
    TS: Time Series of scalar values. The generic form is TS[SCALAR], where SCALAR represents the element type of the 
        time-series. An example of a time-series of int values is TS[int].
    TSL: A list collection of homogeneous time-series values, this has the generic form of TSL[TIME_SERIES_TYPE, SIZE], where 
         TIME_SERIES_TYPE represents the type of the time-series values making up the list and SIZE is an instance of 
         Size, describing the size of the list. To indicate a list of TS[int] with size two, use TSL[TS[int], Size[2]].
    TSB: A named collection of heterogeneous time-series values, this has the generic form of TSB[TS_SCHEMA], where
         TS_SCHEMA represents an instance of a TimeSeriesSchema, this is a dataclass defining the properties of the
         bundle. The types of the properties defined in the dataclass must be TimeSeries types.
    TSD: Time Series Dictionary mapping scalar keys to time-series values, the generic form is TSD[K, V], where K is a 
         scalar type and V is a time-series type, for example TSD[str, TS[int]] represents a dictionary mapping string 
         key to a time-series of int values.
    TSS: Time Series Set of scalar values, the generic form is TSS[SCALAR], where SCALAR represents the element type of
         the elements of the set, TSS[str] represents a set of strings.
    REF: This indicates the node (or graph) expects to receive a reference to a time-series and not the time-series itself.
         The generic form is REF[TIME_SERIES_TYPE], for example REF[TS[int]] indicates a reference to an integer time-series.
    TSW: A time-series window, this supports a sliding window over a time-series. The generic form is 
         TSW[TIME_SERIES_TYPE, WINDOW_SIZE, MIN_WINDOW_SIZE], the MIN_WINDOW_SIZE is optional and defaults to WINDOW_SIZE.
         For example, TSW[TS[int], WindowSize[5], WindowSize[3]] indicates a window of size 5 with a minimum size of 3.
    SCALAR: Represents a non-time-series value. These should be immutable types such as int, float, str, bool, etc.
    TIME_SERIES_TYPE: Represents a time-series type.
    OUT: Represents a time-series value, often used as a return type.
    compute_node: Node decorator indicating the node has both inputs and output time-series values.
    sink_node: Node decorator indicating the node has no inputs and output time-series values.
    generator: Decorator indicating the node is a generator function, this is also a pull source node.
    graph: Used to indicate the wrapped function is a wiring function (describes the relationship between nodes).
    push_queue: Indicates a push source node, the wrapped function is called at start and is provided a sender function
                as an input that can be used to push values to the node.

# You can override these on a per-directory basis using .c7local.yaml files if needed.