Share your --measure or benchmark results

[sravan27]

Show us your measured savings.

After running:

curl -fsSL https://raw.githubusercontent.com/sravan27/context-os/main/setup.sh | bash -s -- --measure

Paste the output below along with:

• Your stack (Next.js, Python, Rust, Go, Flutter, Docker, etc.)
• Approximate repo size (source files + noise files are both printed by --measure)
• Whether you are on Pro, Max, or API

If you want to go further, run the real benchmark:

git clone https://github.com/sravan27/context-os
cd context-os
scripts/benchmark.sh /path/to/your/repo --model sonnet

It parses real claude --print --output-format json usage and reports input/output/cache/cost delta. Paste the JSON report from /tmp/cos-last-benchmark.json and we will add it to the README benchmark list (with your permission).

My fixture (starting point)

On a trivial 2-file fixture:

Metric	Before	After	Delta
Total tokens	79,790	54,036	-32%
Cost (USD)	$0.049	$0.040	-18.8%

On real repos with actual node_modules, the savings get bigger, not smaller.