Skip to main content

JSONL Session Structure Analyzer

Analyzes Claude Code JSONL session files to identify safe split points, message boundaries, and optimal chunking strategies for large sessions.

Features:

  • Stream processing (no full file load)
  • Safe split point detection (file snapshots, user messages, assistant end turns)
  • Tool call sequence tracking (prevents unsafe splits)
  • Overlap window calculation
  • Session metadata extraction

Author: Claude + AZ1.AI License: MIT

File: jsonl_analyzer.py

Classes​

JournalEntryType​

JSONL entry types in Claude Code session files

SplitPointQuality​

Quality/safety level of split points

SplitPoint​

Represents a potential split boundary

SessionStructure​

Session file structure analysis

JSONLAnalyzer​

Analyze JSONL session file structure and find safe split points.

Functions​

analyze_structure(quick_mode)​

Scan file and analyze structure without loading into memory.

find_safe_split_points(target_chunk_size, min_chunk_size, max_split_points)​

Find optimal split points for chunking based on target chunk size.

calculate_overlap(split_line, overlap_messages)​

Calculate overlap window for conversation continuity.

Get recommended chunking strategy for this session.

Usage​

python jsonl_analyzer.py