Token Tracker

Track and report token usage and cost for conversations. Use when: (1) User asks about token consumption (e.g. "token用了多少", "花了多少钱", "how much did this cost"...

Security Scan

The skill is an instruction-only helper that formats session token/cost data from an agent-provided session_status API; it requests no credentials, installs nothing, and stays within its stated purpose.

Purpose & Capability

Name and description match the SKILL.md instructions: it only aims to read session metrics and produce a formatted summary. It does not request unrelated credentials, binaries, or config paths.

Instruction Scope

The instructions tell the agent to call session_status and format results. This stays within scope, but the skill implicitly depends on the platform providing a session_status method that exposes token/cost/cache/context metrics. If the platform exposes additional sensitive data via that API, the skill could surface it — the SKILL.md does not attempt to read files, env vars, or external endpoints.

Install Mechanism

No install spec and no code files — instruction-only skills do not write to disk or fetch remote code, which is lowest-risk for installation.

Credentials

No environment variables, credentials, or config paths are required. The declared requirements are proportionate to the stated purpose.

Persistence & Privilege

Flags are default (always: false, agent can invoke autonomously). The skill does not request permanent presence or modify other skills; no elevated persistence or privileges are requested.

Assessment

This skill is instruction-only and appears coherent: it formats and reports metrics returned by an agent-provided session_status call. Before enabling, confirm your agent platform actually exposes session_status and that the data it returns is what you expect (tokens, cost, cache, context). The skill does not request credentials or install code, but pricing numbers in the SKILL.md are reference-only and may be outdated — verify pricing from your provider if accuracy matters. If you are concerned about privacy, test the skill on non-sensitive conversations to see exactly which fields it surfaces.

Author: zruler
Blog: https://www.zruler.fun/
Email: [e-mail]
Feedback: 有问题请及时反馈！Issues and suggestions welcome.

Report token usage and estimated cost for the current conversation.

When to Trigger

1. User asks — "token用了多少"、"花了多少钱"、"cost"、"usage" 等
2. Long conversation ends — 复杂任务完成后主动报告
3. User requests summary — 明确要求看用量统计

Quick Usage

Call session_status to get current session metrics, then format a user-friendly summary.

Response Format

After getting session_status, report in this format:

📊 本次对话用量

🧮 Tokens: {input} 入 / {output} 出
💵 费用: ${cost}
🗄️ 缓存: {cache_percent}% 命中 ({cached} 缓存 / {new} 新)
📚 上下文: {used}/{max} ({percent}%)

Model Pricing Reference (USD per 1M tokens)

Model	Input	Output	Cached Input
claude-sonnet-4-20250514	$3	$15	$0.30
claude-opus-4-20250514	$15	$75	$1.50
gpt-4o	$2.50	$10	$1.25
gpt-4o-mini	$0.15	$0.60	$0.075
o1	$15	$60	$7.50
o3-mini	$1.10	$4.40	$0.55
gemini-2.0-flash	$0.10	$0.40	$0.025
gemini-2.5-pro	$1.25-$2.50	$10-$15	-
deepseek-chat	$0.27	$1.10	$0.07
deepseek-reasoner	$0.55	$2.19	$0.14

Note: Actual cost from session_status is authoritative. This table is for reference only.

Usage Tips

• For long conversations, suggest compaction if context > 80%
• Note cache hit rate — high cache = lower actual cost
• If user asks for historical usage, explain this tracks current session only