Package index • rrlmgraph

Core API

Primary user-facing functions for building a project graph, querying it for LLM context, and generating static instructions.

build_rrlm_graph(): Build an RLM-Graph for an R project
query_context(): Query the rrlm graph and retrieve a token-budgeted context
chat_with_context(): Chat with an LLM using graph-derived context
generate_instructions(): Generate a Copilot-style instruction file from the project graph
update_graph_incremental(): Incrementally update the graph for changed files

Parse R source files into nodes and edges.

detect_rproject(): Detect an R project type and discover its files
extract_function_nodes(): Extract function nodes from R source files
build_call_edges(): Build CALLS edges between function nodes
build_import_edges(): Build IMPORTS edges from R files to package dependencies
build_test_edges(): Build TEST edges from test files to user-defined functions
build_co_change_edges(): Build CO_CHANGES edges from git commit history
build_dispatch_edges(): Build DISPATCHES_ON and EXTENDS edges for OOP patterns
find_calls_in_body(): Find function calls within an R function body
embed_nodes(): Embed function nodes with TF-IDF, Ollama, or OpenAI vectors
embed_query(): Embed a free-text query into the model's vector space
build_node_context(): Build a text representation of a single graph node
assemble_context_string(): Assemble a structured, LLM-ready context string from ranked hits
compute_relevance(): Compute composite relevance score for a graph node
cosine_similarity(): Compute cosine similarity between two numeric vectors

Save, load, and export the graph.

Record and weight LLM interaction feedback.

Print, summary, and plot for rrlm_graph / rrlm_context.