3.4Side-by-Side Comparison

Metric	MCP (Playwright)	CLI skill	Ratio
Fixed overhead per turn	~5,000 tokens	~50 tokens	~100x cheaper
Per interaction (naive arithmetic)	~10,200 tokens	~130 tokens	~78x cheaper
Real measured task (Microsoft benchmark)	~114,000 tokens	~27,000 tokens	~4x cheaper

Note the gap between the naive per-step arithmetic and the measured end-to-end number, because it's an architect-level insight: in a real task the CLI agent still reads snapshots from disk when it needs them — the savings come from reading state on demand instead of receiving it on every action. The mechanism is "the agent decides what enters context," not "state never enters context." Measured savings: ~4x per task (Microsoft's own benchmark, ~27k vs ~114k tokens), with longer sessions reporting up to 10x — still the difference between finishing a test session with room to reason and compacting context halfway through.