token-check — Claude Code skill: what it does, how to install
Estimates token usage for snippets, files, or chat history. Shows how context limits are consumed and suggests ways to trim the prompt.
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a claude-codeWhat token-check does
The skill provides quick heuristic token counts for files, code snippets, or conversational context. It uses rough character ratios: about 4 characters per token for English prose, 3 for source code, JSON, and logs, and 1.5 to 3 characters for non-Latin alphabets like Cyrillic. Images take about 1,500 tokens, while PDFs combine text length with image costs. Totals are compared against context limits, such as 1M tokens for current Opus, Sonnet, and Haiku 5.5, or 200K on older Haiku builds.
Exact counts require the official API counter. These heuristic figures exist to highlight compounding costs. In ongoing conversations, past attachments are resent on every turn. A 20K token file carried through 30 turns consumes roughly 600K tokens. To counter this, the skill recommends sending only target functions, stripping base64 or repetitive logs, and auditing sessions with /cco audit.
What it helps with
- calculate token estimates for files, logs, or pasted text
- understand compounding context costs across multi-turn chats
- identify unneeded data such as base64 or repetitive logs to cut
- check conversation size against the 1M or 200K context window
Example prompts for the agent
- How many tokens will this log file take and how can I trim it?
- Estimate the token size of our current conversation
- Is this PDF too large to send without hitting limits?
How to install token-check
1Skills CLI (Vercel)
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a claude-code2Manually: put the files in place
~/.claude/skills/token-check/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/token-check/
1Skills CLI (Vercel)
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a codex2Manually: put the files in place
.agents/skills/token-check/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a cursor2Manually: put the files in place
.cursor/skills/token-check/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install egorfedorov/claude-context-optimizer token-check2Manually: put the files in place
.github/skills/token-check/SKILL.md1Skills CLI (Vercel)
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a gemini-cli2Platform CLI
gemini skills install https://github.com/egorfedorov/claude-context-optimizer --path skills/token-checkManually: put the files in place
.devin/skills/token-check/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add egorfedorov/claude-context-optimizer --skill token-check -a opencodeClaude.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 token-check
How to install token-check in Claude Code?
Run npx skills add egorfedorov/claude-context-optimizer --skill token-check -a claude-code. Alternatively, place the file directly at ~/.claude/skills/token-check/SKILL.md.
What does the token-check skill do?
It estimates token counts for pasted snippets, attachments, or conversation histories using character-based heuristics. It also explains context window impact and suggests steps to reduce prompt size.
Are token-check counts exact?
No, they are heuristic ranges because exact counts require an API token counter. For non-Latin scripts like Cyrillic, it calculates 1.5 to 3 characters per token, reflecting higher token consumption than English.
Source text
SKILL.md · MIT
--- name: token-check description: Estimate how many tokens a pasted text, an attached file or the conversation so far takes, what it means for the context window and usage limits, and how to cut it. Use when the user asks how big something is in tokens, why they keep hitting limits, or whether a file is too large to send. license: MIT --- # Token check Give a quick, honest estimate and the cheapest way to get the same result. These are estimates; say so. Exact counts come from the API's token counter. ## Estimate - English prose: about 4 characters per token, roughly 750 words per 1,000 tokens. A full page of text is 500–700 tokens. - Code, JSON, logs: about 3 characters per token. - Cyrillic, CJK and other non-Latin scripts: about 1.5–3 characters per token, so the same text costs noticeably more than in English. - Images: on the order of 1,500 tokens each at typical sizes. PDFs: the text, plus an image's worth for pages read as images. Give a range (for example "about 18–24K tokens"), and compare it with the context window: 1M tokens on current Opus, Sonnet and Haiku 5.5 models, 200K on older Haiku. ## Explain the real cost In a conversation, everything already in it is re-sent with every new message. A 20K-token document pasted early is paid for again on each of the next 30 messages, about 600K tokens in total. That, not the first send, is what uses up limits. ## Ways to cut it, cheapest first 1. Send only the part that matters: the section, the function, the failing lines of a log. 2. Strip what carries no meaning: minified code, base64, repeated log lines, boilerplate. 3. Summarize a long document once, then continue from the summary. 4. When a conversation is long, start fresh from a handoff (the `fresh-start` skill). 5. In Claude Code: read files with offset/limit, let the `context-optimizer` scout subagent do wide exploration on Haiku, and run `/cco audit` to see where a session's tokens actually went.
© egorfedorov · MIT · egorfedorov/claude-context-optimizer
