Scan and merge Claude Code JSONL history into an HTML insights report.
npx skills add https://github.com/ahostbr/liteharness --skill ls-insights-deep
Please help me install this Agent Skill. Skill: ls-insights-deep Source: https://github.com/ahostbr/liteharness/tree/main/liteharness/catalog/skills/ls-insights-deep Command: npx skills add https://github.com/ahostbr/liteharness --skill ls-insights-deep
It eliminates the guesswork of understanding long-term Claude Code usage by producing a comprehensive analysis across the full history of sessions, not just recent activity.
Use case: When you want to understand how your work focus, tooling, friction categories, and outcomes have changed over months of Claude Code sessions, this is the right analysis pass.
Ask for deep insights report generation by scanning the aggregated Claude Code usage dataset and producing the HTML report at the default output path for this skill.
High-intent search queries and answers about installing and using this skill.
To analyze Claude Code JSONL conversation history, this skill scans and merges full-history session data into a structured dataset, quantifying projects, tools, productivity, and friction patterns over time. It aggregates facets and session-meta to provide a comprehensive retrospective analysis.
Yes, you can generate an HTML report from Claude Code sessions by loading a mockup skill to render visualizations of your aggregated usage data. The process synthesizes JSON output from four parallel structured analyses and saves the resulting self-contained report to a file.
Session clustering and friction analysis in AI coding assistants involves grouping historical conversations to identify productivity velocity patterns and quality signals. This skill applies four parallel analysis lenses to extract friction categories, outcomes, and evolution over time from raw JSONL transcript sources.
No external dependencies are required to run JSONL aggregation and deep usage insights generation. The skill operates independently to scan full-history Claude Code data, extract first-prompts and tool counts, and execute parallel structured analyses without needing additional packages.
The best way to quantify project topics and tool usage across long-term AI sessions is running a multi-lens diagnostic breakdown on aggregated JSONL data. This approach extracts tool counts and first-prompts from raw sessions to map the project landscape and track evolution over months.
dparedesiAggregate AI assistant usage data into interactive HTML dashboards.
cogni-workAggregate TIPS project data outputs into a single HTML visualization.
Thiago-Cruz-engAnalyze datasets to extract insights and produce structured reports.
Oruga420Analyze Claude Code JSONL session logs to generate usage analytics and friction insights.
