SuperMarioYL

Ctxprof — Development skill for Claude Code

Development community

Profile Claude Code JSONL sessions into six estimated token buckets, compare runs, and identify large context consumers.

How to install Ctxprof

This entry records only its repository, not the path inside it, so there is no exact command to give. Open SuperMarioYL/ctxprof and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Ctxprof does

Profile Claude Code JSONL sessions into six estimated token buckets, compare runs, and identify large context consumers.

Alternatives in Development

Ideal Customer Profile — Identify ICP with demographics, behaviors, and JTBD 7.8k ★
Om Standup — Morning kickoff 4.6k ★
Claude Supermemory — Persistent memory across sessions and projects using Supermemory 2.3k ★

README

[简体中文](./README.zh-CN.md) · [Website](https://ctxprof.lei6393.com) · [GitHub](https://github.com/SuperMarioYL/ctxprof)

ctxprof

**See what occupies your context window.**

ctxprof reads Claude Code session logs and attributes recorded usage to system, skill, MCP, file, reasoning and output buckets.

Why use it

An aggregate usage number cannot tell you which loaded content deserves attention. A ranked allocation helps you inspect the largest consumers and compare sessions before changing your workflow.

Six useful buckets — Inspect system, skills, MCP, files, reasoning and output.
Separate peak from throughput — Cached prefixes are not treated as fresh window capacity on every turn.
Compare before trimming — Trend and compare operate on explicit session files.

Architecture

The parser reads JSONL turns, a vendored byte-level BPE weights individual blocks, and reconciliation scales those weights to recorded usage. Attribution groups the result into six buckets. The renderer separates peak single-turn window occupancy from cumulative throughput.