receipts · karanb192 — Claude Code skill: model quality trends
receipts — a skill that displays your personal quality card across sessions: failure rate, edit churn, and tokens per task broken down by model version. Runs silently in the background and surfaces shifts in model performance.
MIT ★ 586 updated 2026-10-26
/plugin marketplace add karanb192/claude-code-hooks
/plugin install nerf-receipts@claude-code-hooksWhat receipts does
receipts aggregates metrics during Claude Code sessions and renders a card with three per-session measures: failure rate (proportion of failed attempts), edit churn (how much code was rewritten), and tokens per task. Each metric includes a sparkline showing the trend across recent sessions. The card is keyed by model id and Claude Code version, so you can spot if quality changed after an update.
When you invoke the skill, it interprets the card for you: calls out headline averages and, if a ⚠ warning line appears, restates which metric moved, by how much, and across which model change. If data is still accumulating, it suggests you keep working and check back after a few more sessions.
What it helps with
- track model output quality across your working sessions
- detect a drop in quality after a Claude Code update
- compare metrics between different model versions
- identify if code churn is increasing over time
Example prompts for the agent
- Run receipts to display your quality card with all metrics
- Check failure rate and edit churn trends from recent sessions
- Compare performance between model versions in the card
How to install receipts
1Plugin marketplace
/plugin marketplace add karanb192/claude-code-hooks/plugin install nerf-receipts@claude-code-hooksThis installs the whole nerf-receipts plugin with all its skills
2Skills CLI (Vercel)
npx skills add karanb192/claude-code-hooks --skill receipts -a claude-code3Manually: put the files in place
~/.claude/skills/receipts/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/receipts/
1Skills CLI (Vercel)
npx skills add karanb192/claude-code-hooks --skill receipts -a codex2Manually: put the files in place
.agents/skills/receipts/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add karanb192/claude-code-hooks --skill receipts -a cursor2Manually: put the files in place
.cursor/skills/receipts/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install karanb192/claude-code-hooks plugins/nerf-receipts/skills/receipts2Manually: put the files in place
.github/skills/receipts/SKILL.md1Skills CLI (Vercel)
npx skills add karanb192/claude-code-hooks --skill receipts -a gemini-cli2Platform CLI
gemini skills install https://github.com/karanb192/claude-code-hooks --path plugins/nerf-receipts/skills/receiptsManually: put the files in place
.devin/skills/receipts/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about receipts
How do I install receipts in Claude Code?
Run /plugin marketplace add karanb192/claude-code-hooks, then /plugin install nerf-receipts@claude-code-hooks. You can also install manually with npx skills add karanb192/claude-code-hooks --skill receipts -a claude-code.
What metrics does the receipts card show?
Three metrics with sparklines: failure rate (proportion of failed attempts in your sessions), edit churn (how often code was rewritten), and tokens per task (average token spend). Data is grouped by model id and Claude Code version so you can spot performance shifts.
When does data start appearing in the receipts card?
The skill collects metrics in the background starting from your first session. The full card populates after a few sessions; if data is still sparse, the skill will tell you to keep working and check back later.
Source text
SKILL.md · MIT
---
name: receipts
description: Show your personal model-quality trend card (failure rate, edit churn, tokens/task by model version)
disable-model-invocation: true
---
# Nerf receipts: your model-quality trend card
!`node "${CLAUDE_SKILL_DIR}/../../nerf-receipts.js" --render 2>/dev/null || node "${CLAUDE_PLUGIN_ROOT}/nerf-receipts.js" --render`
The card above is your own per-session flight recorder, aggregated across your recent
Claude Code sessions and keyed by model id and Claude Code version: failure rate, edit
churn, and tokens per task, each with a sparkline. It is recorded quietly in the background
by the `nerf-receipts` hooks as you work.
Briefly interpret it for the user: call out the headline averages, and if a ⚠ shift line is
present, restate which metric moved, by how much, and across which model change; that is the
receipt that a quality shift is real, not a vibe. If the card says data is still accumulating,
tell them to keep working and check back after a few more sessions. Keep it short and do not
re-run any commands.
© Karan Bansal · MIT · karanb192/claude-code-hooks
Similar skills
All skills for Agent setup: memory, context, orchestration →receipts by other authors comes first.