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.
npx agentmods add instructions/kendrick-na/tokenjuice/agents-mdgit clone --depth 1 https://github.com/kendrick-na/tokenjuiceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/kendrick-na/tokenjuice/agents-md)<a href="https://agentmods.dev/instructions/kendrick-na/tokenjuice/agents-md"><img src="https://agentmods.dev/badge/instructions/kendrick-na/tokenjuice/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.02066 | $0.02066 |
| Opus 5 | $0.01033 | $0.01033 |
| Sonnet 5 | $0.00413 | $0.00413 |
| Haiku 4.5 | $0.00207 | $0.00207 |
Grade A, and why
tokenjuice AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 184 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — installing tokenjuice
Instructions for a coding agent (Claude Code, Codex, Cursor, …) asked to install this for a user. Follow the branch for the user's OS; do not run both.
What this is: a menu bar / system tray battery showing how much Claude Code
and Codex usage the user has left. macOS uses SwiftBar; Windows uses a Python
tray app. Both run the same engine, claude-codex-battery.5s.js.
Step 0 — is it already installed?
Check before cloning. Cloning blindly next to an existing copy leaves two checkouts and the user can no longer tell which one is live.
ls ~/.swiftbar-plugins/claude-codex-battery.5s.js # macOS: installed?
ls "$(cygpath -u "$USERPROFILE" 2>/dev/null)/AppData/Roaming/Microsoft/Windows/Start Menu/Programs/Startup/tokenjuice.lnk" # Windows
If either exists, this is an update, not an install. Find the existing clone
(git -C <path> remote -v should show tokenjuice) and re-run the installer
there — both are idempotent and re-copy the current engine. Do not clone a
second copy.
git pull first only if the checkout is clean. Check git status — a user who
edited files locally should be asked before you touch their work.
Step 1 — detect the OS
Do not assume. Windows and macOS have entirely separate installers, and running the wrong one wastes the user's time.
| Signal | OS | Go to |
|---|---|---|
uname prints Darwin |
macOS | macOS |
$env:OS is Windows_NT, or you're in PowerShell/cmd |
Windows | Windows |
uname prints Linux |
Linux | Linux |
If the user is on Windows but you only have a bash shell, check whether it's WSL
(uname -r contains microsoft). WSL is the Linux path, not the Windows one —
a tray icon from inside WSL won't reach the Windows taskbar, and WSL has its own
~/.claude, so it reports the usage of the Linux-side Claude Code, not Windows'.
Ask which one the user means before continuing.
macOS
Ask the user where to clone if they have a convention; otherwise clone into a directory you know, not whatever the shell happens to be sitting in:
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.
- 5d ago First seen · 184 lines · 2,066 tokens per session scan A 2e72d486961a
tokenjuice AGENTS.md is an instructions file published in the GitHub repository kendrick-na/tokenjuice (2 stars, last pushed 9d ago), licensed MIT. It adds 2,066 tokens to every session, about $0.0103 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.
Other instructions, from other repositories
nico-apple-mcp AGENTS.md
AGENTS.md instructions for proxynico/nico-apple-mcp, covering agents.md, cursor cloud specific instructions, prerequisites, key commands and linux / cloud agent caveats.
nico-apple-mcp CLAUDE.md
Claude Code instructions for proxynico/nico-apple-mcp: Bun/TS MCP server for Apple Calendar, Notes, and Contacts. Commands and Linux/cloud caveats live in AGENTS.md.
opengeni AGENTS.md
AGENTS.md instructions for Cloudgeni-ai/opengeni, covering agent / automation notes (opengeni), full local stack, architecture notes, pull-request delivery across moving main and keeping these notes current.
starboard CLAUDE.md
Claude Code instructions for palamim/starboard, covering claude.md, commands, code style, architecture and non-obvious gotchas (not discoverable by reading the code).
AutoControlGUI CLAUDE.md
Claude Code instructions for Integration-Automation/AutoControlGUI, covering claude.md — autocontrol, project overview, architecture, development commands and feature delivery rules.
cctop CLAUDE.md
Claude Code instructions for st0012/cctop: This repository keeps shared agent instructions in AGENTS.md. Read AGENTS.md now and follow it as the source of truth for cctop development.