Free tool
Token & Context Estimator
Estimate how many tokens a prompt or SKILL.md uses, and how
much of a model's context window it fills. Everything runs in your browser — nothing is
sent to any server.
Estimated range
— tokens
Context window usage
This is a rough estimate (~4 characters per token). Actual token counts vary by model and tokenizer — use it for planning, not exact billing.
Why token count matters for skills
Every skill your agent loads consumes context. Keeping skills focused leaves more room for the actual task — and keeps latency and cost down.
Context budget
A model's context window is finite. Skills, tool outputs, and history all share it — trim skills so the task has room.
Cost & latency
More input tokens mean higher cost and slower responses. Concise skills are cheaper to run at scale.
Load on demand
With MCP, an agent fetches a skill only when it needs it — so keep each skill tight and single-purpose.
Keep your team's skills lean and governed
SkillInvoke stores skills in Git, reviews changes by PR, and serves them on demand over MCP — so agents load only what they need.