graph

name	graph
description	Manage code knowledge graphs via code-review-graph (CRG). Build, update, and check status of project code graphs stored in .code-review-graph/graph.db. Use when: (1) user says "build graph", "update graph", "graph status", "/graph", (2) Harness init detects CRG, (3) preparing to use /explore commands. Gracefully degrades if CRG is not installed.

Graph — Knowledge Graph Management

Overview

Manage project code knowledge graphs powered by code-review-graph. The graph stores code structure (functions, classes, calls, imports) in .code-review-graph/graph.db (SQLite) and enables architecture understanding, impact analysis, and caller/callee tracing.

Announce at start: "I'm using the graph skill to manage the project's code knowledge graph."

Commands

/graph build — Full Build

Build a complete code graph for the current project:

# Check CRG is available first
python -c "import code_review_graph.tools" 2>/dev/null
if [ $? -ne 0 ]; then
    echo "CRG not installed. Run: pip install code-review-graph"
    exit 1
fi

# Full build
code-review-graph build --repo .

After build, report:

• Graph location: .code-review-graph/graph.db
• Node count: SELECT COUNT(*) FROM nodes
• Edge count: SELECT COUNT(*) FROM edges
• Community count: use list_communities_func from code_review_graph.tools

/graph update — Incremental Update

Update the graph with recent changes (faster than full build):

code-review-graph update --repo .

/graph status — Graph Status

Check if a graph exists and show stats:

import os, sqlite3

db_path = ".code-review-graph/graph.db"
if not os.path.exists(db_path):
    print("No graph found. Run /graph build first.")
else:
    conn = sqlite3.connect(db_path)
    nodes = conn.execute("SELECT COUNT(*) FROM nodes").fetchone()[0]
    edges = conn.execute("SELECT COUNT(*) FROM edges").fetchone()[0]
    mtime = os.path.getmtime(db_path)
    from datetime import datetime
    last_updated = datetime.fromtimestamp(mtime).strftime("%Y-%m-%d %H:%M")
    print(f"Graph: {db_path}")
    print(f"Nodes: {nodes} | Edges: {edges} | Last updated: {last_updated}")
    conn.close()

Also run list_communities_func from code_review_graph.tools to show community count.

/graph view — Interactive Visualization in Browser

Open an interactive graph visualization in the browser:

# Generate HTML + start local server (auto-opens http://localhost:8765)
code-review-graph visualize --repo . --serve

View modes:

# Default: auto-detect best layout
code-review-graph visualize --repo . --serve

# Community/module view: nodes grouped by detected communities
code-review-graph visualize --repo . --mode community --serve

# File-level view: nodes are files, edges are cross-file dependencies
code-review-graph visualize --repo . --mode file --serve

# Full view: every function/class as a node
code-review-graph visualize --repo . --mode full --serve

Export (no server):

# Generate HTML file only → open .code-review-graph/visualization.html manually
code-review-graph visualize --repo .

# Export as SVG image
code-review-graph visualize --repo . --format svg

# Export for Obsidian
code-review-graph visualize --repo . --format obsidian

# Export as GraphML (for yEd/Gephi)
code-review-graph visualize --repo . --format graphml

/graph wiki — Generate Markdown Wiki

code-review-graph wiki --repo .
# Generates markdown docs from community structure

/graph install — Check/Install CRG

python -c "import code_review_graph.tools; print('CRG is available')" 2>/dev/null \
  || echo "CRG not installed. Install with: pip install code-review-graph"

Graceful Degradation

• CRG not installed → show install command, never crash
• Detection method: python -c "import code_review_graph.tools" (matches graph_builder.py pattern)
• Graph file missing → prompt /graph build

Storage

• Graph: .code-review-graph/graph.db (CRG default, in workspace)
• Metadata: .code-review-graph/ directory, managed by CRG automatically