Quality record

Know what you are installing.

Version, date and limits are read from the repo on every sync, and the check runs every night. Real-world runs are listed only with their evidence.

Version: Unversionedcommit a4d5e80
Updated: 29 Oct 2026last commit to the repo
Works in: Claude Code, Codex, Cursor, Qoder and any agent that loads a SKILL.mdplain text; no API keys
Nightly check: Not checked yetIt joins the nightly check after its first run. How it is checked
Last real-world run: No public run yetRuns are published only with their evidence.
Known limits: Not built for:
Choosing an effort level before work starts or using less going forward (use choose-effort)
For plan limits, prices and API billing (use product-self-knowledge)

What it is for

Diagnoses why a Claude Code session or day burned through usage limits faster than expected, and says what to change. Reads the user's own /usage breakdown, checks the known heavy-session causes (high effort levels, subagents inheriting effort, goal mode, workflows, agent teams, long SDK jobs), and returns up to three ranked culprits with one fix each, plus when to send /feedback.

Reach for it when

The user says Claude Code usage is draining fast, they hit or are nearing their limit sooner than expected, one session cost far more than expected, or asks what /usage shows or how to cut burn in a session that already ran.

Not for

Choosing an effort level before work starts or using less going forward (use choose-effort)
For plan limits, prices and API billing (use product-self-knowledge)

Every file it installs

Measured from the packed archive at commit a4d5e80. A skill is plain text your agent reads, so this list is all of it.

File	Lines	Bytes
SKILL.md	34	3,441 B
references/burn-sources.md	30	2,849 B
references/trigger-cases.md	21	2,247 B
3 files	85	4 KB

Diagnose Usage

Know what you are installing.

What it is for

Reach for it when

Not for

Every file it installs

More in Build & Ship