tree-sitter-extraction

name	tree-sitter-extraction
description	Extract symbols, imports, and call edges from source code using tree-sitter for codebase indexing MCP workflows. Use when implementing parser-based code intelligence.

Tree-sitter Extraction

When to Use

• Build AST-based extraction for dependencies/call chains.
• Replace fragile regex-based code analysis.

Procedure

1. Select supported languages and parser versions.
2. Define extraction contract per language:
   • file-level module id
   • symbol definitions
   • import/export relations
   • invocation/call edges
3. Normalize extracted entities into common graph schema.
4. Track unresolved symbols separately (do not silently drop).
5. Add parser failure counters and fallback strategy.

Quality Checks

• Stable symbol IDs across incremental runs.
• Deterministic extraction for unchanged files.
• Bounded parse time and memory usage.