Code graph analysis

BuddhiAI CLI parses supported source files with tree-sitter and records code entities and their relationships. Run buddhi init for the first graph, then use buddhi generate to rebuild it after code changes.

The generated .buddhi/graphs/ directory contains:

• tree-graph.db, a SQLite database used for graph queries.
• tree-graph.json, a Cytoscape-compatible representation.
• tree-graph.html, an interactive graph viewer. It loads Cytoscape.js from a CDN, so the browser needs internet access to load that dependency.

Inspect graph structure

Run these commands from the repository root:

buddhi graph stats
buddhi graph analyze
buddhi graph community
buddhi graph cycles

• stats reports graph counts, confidence breakdown, and Git staleness.
• analyze reports highly connected nodes, surprising connections, and review questions.
• community inspects graph clustering and cohesion.
• cycles finds circular import chains.

Trace dependencies

buddhi graph path MyService AuthManager
buddhi graph blast-radius

path finds a shortest graph path between symbols and can disambiguate names. blast-radius analyzes transitive callers and affected communities to help assess change impact.

Run buddhi graph --help for the options supported by your installed version.

Keep generated data current

buddhi generate [path] rebuilds all graph artifacts under the target repository. It does not update .buddhi/docs-plan.json; use buddhi docs plan for that separate step.

Graph resolution is best-effort, not a complete semantic analysis. The CLI resolves same-project relative imports and selected same-file calls; external or ambiguous relationships may remain unresolved.