code-graph-ingest

name	code-graph-ingest
description	Parse and ingest source code into Neo4j and Qdrant using the Python analyzers in graph-code-tiny. Use when asked to scan repositories, build call graphs, store symbols/relations in Neo4j, push embeddings into Qdrant, tune batching/caching, or run Kotlin/Java/C#/C++/TypeScript/JavaScript/PHP/SQL/PL/SQL analyzers.

Code Graph Ingest (graph-code-tiny)

Workflow

1. Identify language(s) and root folder(s) to scan.
2. Ensure Python deps are installed from requirements.txt and Tree-sitter is available.
3. Decide where to write:
   • Neo4j only, Qdrant only, or both.
   • Set connection env vars or pass CLI flags.
4. Optionally run a dry-run to count files.
5. Run the language-specific analyzer.
6. Validate results (Neo4j writes + Qdrant collection updated).

Analyzer selection

• Kotlin: tools/kotlin/kotlin_analyzer.py
• Java: tools/java/java_analyzer.py
• TypeScript: tools/ts/ts_analyzer.py
• JavaScript: tools/js/js_analyzer.py
• PHP: tools/php/php_analyzer.py
• SQL: tools/sql/sql_analyzer.py
• PL/SQL: tools/plsql/plsql_analyzer.py
• C#: tools/csharp/csharp_analyzer.py
• C/C++: tools/cplus/cplus_analyzer.py

Key behaviors to remember

• Neo4j writes happen only when --neo4j-* credentials are provided.
• Qdrant writes happen only when --qdrant-url is provided.
• Parsers support caching and resume; disable explicitly if needed.
• C/C++ analyzer supports extra export files (event map, call stats, possible/unresolved calls).
• Kotlin uses embedder vector size from the model; Java/C#/C++ assume 768.

References

• references/env.md for environment variables and defaults.
• references/analyzers.md for full CLI flags and per-language specifics.
• references/examples.md for common command templates.