build-graph

Build Knowledge Graph

Build a knowledge graph from the current directory, then embed all nodes for semantic search across sessions.

Check Engine Availability

python -c "
import sys, os
sys.path.insert(0, os.environ.get('CLAUDE_PLUGIN_ROOT', '.'))
from tentaqles.graph import get_engine
try:
    engine = get_engine()
    print(f'Engine: {engine.name}')
except RuntimeError as e:
    print(f'ERROR: {e}')
"

If no engine is available, tell the user their options:

• pip install graphifyy — uses the graphify package (recommended, most mature)
• pip install tentaqles[graph] — uses the native Tentaqles engine (built-in, includes all enhancements)

Build the Graph

If using graphify engine: Run /graphify on the current directory. The graphify skill handles the full pipeline (detection, AST extraction, semantic extraction with subagents, clustering, reporting, HTML visualization). After it completes, continue to the embedding step below.

If using native engine:

python -c "
import sys, os
sys.path.insert(0, os.environ.get('CLAUDE_PLUGIN_ROOT', '.'))
from tentaqles.graph import get_engine
from pathlib import Path

engine = get_engine()
result = engine.build(Path(os.getcwd()))
print(f'Graph built: {result.get(\"nodes\", 0)} nodes, {result.get(\"edges\", 0)} edges, {result.get(\"communities\", 0)} communities')
print(f'Output: {result.get(\"output_dir\", \"graphify-out/\")}')
"

Embed Nodes for Semantic Search

After the graph is built (by either engine), embed all nodes:

python -c "
import sys, os
sys.path.insert(0, os.environ.get('CLAUDE_PLUGIN_ROOT', '.'))
from tentaqles.graph import get_engine

engine = get_engine()
result = engine.embed('graphify-out/graph.json')
print(f'Embedded {result[\"nodes_embedded\"]} nodes ({result[\"file_size_kb\"]} KB)')
print('Semantic search is now active for /tentaqles:query-memory')
"

Report

Tell the user:

• How many nodes, edges, communities were found
• Where the outputs are (graphify-out/)
• That semantic search is now active
• Suggest: "Try /tentaqles:query-memory to search the graph by meaning"
• If GRAPH_REPORT.md exists, highlight the God Nodes and Surprising Connections sections