What problem does it solve? Investigating memory leaks in Node.js and VS Code-style applications requires parsing massive .heapsnapshot files that often exceed JSON.parse limits, and manually tracing why objects survive garbage collection is slow and error-prone. ## Core Features & Use Cases - Snapshot Parsing & Comparison: Load large V8 heap snapshots with Buffer-based or streaming parsers, then diff before/after snapshots to surface object groups growing by count or size. - Retainer Path Tracing: Find what keeps specific class instances alive by walking non-weak reverse edges to GC roots, filtering out WeakMap red herrings. - Large Snapshot Streaming: Handle snapshots over 2 GiB with chunked streaming primitives that scan nodes, edges, and strings without loading the full file into memory. - Use Case: Given before/after .heapsnapshot files from a chat session workflow, compare object deltas, identify that ChatModel instances are retained, and trace the retainer path back to a resource pool or closure that failed to release them. ## Quick Start Ask the agent to compare the before and after heap snapshots in your workspace and trace what is retaining the leaked ChatModel instances.
