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 skills/codeshux/tokenwise/installnpx skills add CodeShuX/tokenwise --skill installgit clone --depth 1 https://github.com/CodeShuX/tokenwiseWhat 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 | $0.00095 | $0.02803 |
| Opus 5 | $0.00048 | $0.01401 |
| Sonnet 5 | $0.00019 | $0.00561 |
| Haiku 4.5 | $0.00010 | $0.00280 |
Grade B, and why
install scanned grade B with 1 finding 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- `~/.claude/settings.json` — note "missing" if absent How it starts
The opening of the file, as written. The whole thing — 241 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/tokenwise:install — Configure model routing
You are the TokenWise installer. Walk the user through Phase 1 (Detect) and Phase 2 (Configure) of the TokenWise lifecycle.
Parse $ARGUMENTS
$ARGUMENTS may contain:
--guided— force guided mode (default if not detected as power user)--manual— print copy-paste blocks, exit without writing--dry-run— show all proposed changes without writing OR printing copy-paste
If empty or contains --guided, run guided mode. If --manual, run manual mode. If --dry-run, run guided flow but skip every write step — print "DRY RUN: would write to " instead.
Phase 1 — Detect
Print a header:
TokenWise Install — Phase 1: Detect
====================================
Do these checks (use Read and Bash tools):
-
Read existing configs:
~/.claude/CLAUDE.md(global) — note "missing" if absent./CLAUDE.md(project, if cwd is a git repo) — note "missing" if absent~/.claude/settings.json— note "missing" if absent
-
Claude Code version: run
claude --versionvia Bash. Record output. -
Subagent routing probe (Anthropic Issue #47488 regression test):
- Spawn a probe Task at Haiku tier:
Task(description: "probe", subagent_type: "general-purpose", model: "haiku", prompt: "Return only the string TOKENWISE_PROBE_OK") - If the response contains
TOKENWISE_PROBE_OK, routing works. If not, mark probe as FAILED. - Note: if your Task tool doesn't support
model:, or you have no Task tool available at all in this session, mark as "routing probe: unverifiable on this build" and proceed the same way — this is not a FAILED result and doesn't block install.
- Spawn a probe Task at Haiku tier:
-
Env-var probe (Anthropic Issue #36381):
- Run
bash -c 'CLAUDE_AUTOCOMPACT_PCT_OVERRIDE=80 env | grep CLAUDE_AUTOCOMPACT' - If output contains the override, env vars are honored. Note as "env-var probe: OK".
- Run
-
Inventory existing skills/hooks by listing
~/.claude/plugins/cache/*/directories — note any that look like routing or token-tracking tools.
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.
- yesterday First seen · 241 lines · 95 tokens per session scan B 0eac95277aa8
install is a skill published in the GitHub repository CodeShuX/tokenwise (3 stars, last pushed 6d ago), licensed MIT. It adds 95 tokens to every session and 2,803 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
archify
Create polished, validated architecture, workflow, sequence, data-flow, and lifecycle/state diagrams as explorable standalone HTML with inline SVG, dark/light themes, optional trace motion, and PNG/JPEG/WebP/SVG/WebM export. Accept plain-language requirements or pasted Mermaid flowchart, sequenceDiagram, and…
vox-director
Turn ONE topic into a finished Vox-style paper-collage explainer / ad video, end to end on the Atlas Cloud API + local ffmpeg — script, collage keyframes, motion, voice-over, music, captions, all automated. Use this whenever the user wants a "Vox style" video, a paper/torn-paper collage animation, a "motion collage"…
social-post
學習使用者的 Facebook／Instagram／YouTube／Threads／X 語氣與受眾，規劃、撰寫、確認後發佈內容；以已登入 Chrome 受控掃描、草擬及回覆 FB／IG／Threads 留言；並作為流量、留存與轉化的結構化帳本。使用者說「發文」「用我的口氣」「回覆留言」「自動回留言」「掃留言」「查流量」「演算法」「把數據訓練進去」「比較貼文」「優化 pattern」時使用。.
privacy-jp
日本サービス向け（일본 서비스용） プライバシーポリシー・利用規約・同意モーダル・Cookieバナー 자동 생성. 個人情報保護法(APPI)·消費者契約法·特定商取引法 반영. Next.js 1316 프로젝트 대상.
higgsfield
Use this skill whenever the user asks anything about Higgsfield AI — writing or refining video/image prompts, choosing a model (Kling, Sora 2, Veo, Wan, Seedance, Minimax Hailuo, DoP, Soul, Nano Banana, Seedream, Flux, GPT Image, etc.), camera controls, named motion presets, Soul ID character consistency, Cinema…
higgsfield-camera
Use when the user asks about camera movements, shot types, or how to describe camera behavior in a Higgsfield prompt. Contains all named camera controls with descriptions, best use cases, and example prompt phrases.