@royalsolution/ontosight

Visualize CodeGraph call subgraphs in OntoSight without installing Hyper-Extract.

This npm package is a thin wrapper that auto-creates the CodeGraph index when missing, then runs the Python package ontosight-codegraph via uvx (or pipx as fallback).

Prerequisites

• Node 20+ (for npx)
• Python 3.11+ with uv (recommended) or pipx

Quick start

# Opens OntoSight; runs `npx @colbymchenry/codegraph init -i` first if needed
npx @royalsolution/ontosight .

# Seed around a symbol
npx @royalsolution/ontosight . --symbol view_graph --path vendor/ontosight/

# Task-scoped subgraph
npx @royalsolution/ontosight . --task "auth flow" --hops 2 --max-nodes 200

Set ONTOSIGHT_SKIP_AUTO_INIT=1 to require a pre-built index instead of auto-init.

Options

| Flag | Description | |------|-------------| | --path <prefix> | Limit symbols to files under this path | | --symbol <name> | Seed symbol for subgraph expansion | | --task <text> | Natural-language task seed | | --hops <n> | BFS hop depth (default: 2) | | --max-nodes <n> | Maximum nodes (default: 200) |

AI agents

OntoSight complements CodeGraph MCP — use codegraph_* tools to answer structural questions in chat; use this CLI when the user wants an interactive call-graph UI.

| Need | Tool | |------|------| | Find symbols, callers, traces, impact (text) | CodeGraph MCP (codegraph_search, codegraph_context, …) | | Visual exploration in browser | npx @royalsolution/ontosight |

Full agent guide: AGENTS.md — workflows, flags, env vars, troubleshooting.

Project wiring (Hyper-Extract monorepo):

| Tool | Location | |------|----------| | Cursor | .cursor/rules/ontosight.mdc, .cursor/references/ontosight.md | | Kiro | .kiro/steering/ontosight.md, .kiro/references/ontosight.md | | Antigravity | .agent/rules/ontosight.md, .agents/references/ontosight.md |

Example agent workflow

1. codegraph_context({ task: "auth flow", maxNodes: 20 })  → answer in chat
2. npx @royalsolution/ontosight . --task "auth flow" --hops 2   → open graph for user

Hyper-Extract users

If you already have Hyper-Extract installed, the equivalent command is:

he show . --codegraph
he show . --codegraph --symbol view_graph --path vendor/ontosight/

Publish (maintainers)

See PUBLISH.md. Keep ontosightCodegraphVersion in package.json in sync with the PyPI ontosight-codegraph release.