cache

A guide for explaining Claude billing data, especially how cached input is priced. Cached input is previously provided information reused by the model, and the guide covers cache reads, cache writes, effective costs, and savings.

In plain words
What is it for?
Use it to inspect cache usage, calculate cache-related costs and savings from the billing data, and explain why a Claude bill has its recorded amount.
Why use it?
It helps separate the different token charges in a bill instead of treating all input as one cost. The supplied commands retrieve the metered data used for the explanation.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/jakeintech/waybill/cache
Any agent
npx skills add Jakeintech/waybill --skill cache
Clone the repo
git clone --depth 1 https://github.com/Jakeintech/waybill

Made for: Claude Code, Codex.

Per session 112 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 934 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00112 $0.00934
Opus 5 $0.00056 $0.00467
Sonnet 5 $0.00022 $0.00187
Haiku 4.5 $0.00011 $0.00093

Measured yesterday against content hash 644d346781b9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cache scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/cache/SKILL.md · 76 lines

How it starts

The opening of the file, as written. The whole thing — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cache

Most of a Claude Code bill is not what people picture. Cache reads are billed at a tenth of the input rate and usually dominate the volume; cache writes carry premiums (1.25× for 5-minute, 2× for 1-hour entries). This skill answers "why is my bill like this" from the metered ledger. The engine computes every number; you write the prose around them and never adjust a figure.

"${CLAUDE_PLUGIN_ROOT}/bin/waybill" mine --all    # catch-up metering first, always
"${CLAUDE_PLUGIN_ROOT}/bin/waybill" query cache [--from <date>] [--to <date>]

The payload (data):

  • tokens — volume by class: input, output, cache_read, cache_creation (with the 5m/1h write split); cache_read_pct is reads as a share of total volume.
  • by_model — the same per model, each with a derived effective_usd (null when no rate resolves — never $0).
  • billed — the cost view, derived from the current rate table and labeled so (basis: "list_price_equivalent_derived"): effective_usd (the tokens priced as billed), list_equivalent_usd (the same tokens with no caching at all), saved_usd (the NET saving, write premiums already paid for), cache_read_share_of_billed_pct, and covered_pct (how much of the volume the dollar math covers).
  • unattributed_pct — the honesty floor; it travels on every payload.

Rendering rules

Honor the detail axis (config detail_default, --detail override):

  • terse (1–3 lines): total volume, the cache-read share, and the net saving with its basis label. Even at terse, never drop the unattributed %, covered_pct when below 100, or the derived-basis label — those survive every level.
  • standard: the above plus the tier table (input / output / reads / 5m writes / 1h writes) and the effective-vs-list pair.
  • full: everything, including per-model rows.

Language rules (methodology): "saved" is always derived, list-price equivalent — say so; it is not a stored fact and not a promise about the actual invoice. When covered_pct < 100, name the unpriced models (waybill status lists them with fixes) rather than presenting a partial dollar figure as complete. Ranges stay ranges; the unattributed share is shown, never hidden.

Read the full file on GitHub · 76 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 76 lines · 112 tokens per session scan A 644d346781b9

Subscribe to this mod's changes

cache is a skill published in the GitHub repository Jakeintech/waybill (1 stars, last pushed 9d ago), licensed MIT. It adds 112 tokens to every session and 934 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

ideogram-ultra

Build Ideogram 4 (Ideogram Ultra) txt2img and img2img workflows with the local open-weights model, dual conditional/unconditional models with DualModelGuider, Qwen3-VL text encoder, and structured JSON ("compositional deconstruction") prompts for strong text rendering and layout control.

artokun/comfyui-mcp · 68 tokens

model-compatibility

Model family compatibility matrix covering loaders, resolutions, samplers, CFG, VAE, ControlNet, and LoRA compatibility for SD 1.5, SDXL, Flux, SD3, and video models.

artokun/comfyui-mcp · 47 tokens

flux-txt2img

Build Flux txt2img workflows with Flux.1 Dev (SRPO), Flux 2 Klein 9B, Turbo LoRAs, FluxGuidance, and DualCLIPLoader patterns.

artokun/comfyui-mcp · 44 tokens

panel-node-pack-sync

Keep the ComfyUI sidebar panel node-pack (comfyui-agent-panel) in step with the orchestrator after comfyui-mcp updates. Use this whenever the orchestrator was just updated (installcomfyui (action:"selfupdate"), npm i -g comfyui-mcp, a new version in the ENVIRONMENT line), when a panel/bridge command fails in a way…

artokun/comfyui-mcp · 206 tokens

civitai

Discover Civitai models with the BUILT-IN downloadmodel action:"searchcivitai" and install/generate them locally. Find a checkpoint/LoRA/embedding on Civitai, download it into ComfyUI, and use its trigger words. Optionally pair the official Civitai MCP for community features (images browsing, posting, collections).

artokun/comfyui-mcp · 76 tokens

31-offer-design-global

Use when the WHAT being sold needs work, not the price and not the copy — core promise, value stack, bonuses, guarantee and risk reversal, honest scarcity and urgency, tier packaging, upsell and downsell paths, and high-ticket or productized service structures. Trigger on 'design an offer', 'irresistible offer'…

minhnv0807/ai-business-skills · 172 tokens