Atlas › 01 Agent Skills for Browser Automation › Token Budget Analysis: Real Numbers☰ Read as one page
Token Budget Analysis: Real Numbers
8.1Context Windows (July 2026)As of July 2026 the frontier models — Claude Opus 4.8, OpenAI GPT-5.5, Google Gemini 3.1 Pro — ship context windows ranging from the…8.2The Headline Number FirstMicrosoft's published benchmark for a typical browser automation task:8.3MCP Token Breakdown: A Real Playwright SessionAn MCP server exposes 15-25 browser tools, each with a JSON schema (~150-280 tokens apiece). Total: roughly 3,000-4,000 tokens of schemas…8.4Skill Token Breakdown: The Same Test via the Playwright CLINote what the agent did not pay for: seventeen of the twenty steps returned a ref confirmation or a file path costing dozens of tokens. The…8.5Side-by-Side Summary8.6What You Do with the Saved ContextThe ~87K tokens saved on a typical task can hold:8.7When Token Cost Doesn't MatterIf you're on a 1M+ context window (Gemini 3.1 Pro class), the context-exhaustion argument is much weaker. You can afford MCP's overhead and…8.8Interview Talking Point"I anchor the skills-vs-MCP token argument to Microsoft's own benchmark: about 114,000 tokens for a typical browser task through Playwright…