Show HN: Lowfat – pluggable CLI filter that saved 91.8% of my LLM tokens
91.8% token savings. That's what one developer achieved by filtering CLI noise before it hits LLM agents.
Why it matters
As AI agents become production infrastructure, token efficiency at the infrastructure layer—not just model optimization—is becoming a competitive advantage. This tool demonstrates a practical approach to reducing LLM costs through intelligent output filtering in agent workflows.
The key facts
14 to know91.8% total token savings across 2 months of personal use (4.4M raw tokens → 4.1M saved)
kubectl get commands: 93.9% savings on 101 runs
docker commands: 96.1% savings across 127 runs
grep: 96.2% savings on 103 runs
Pluggable CLI filter designed for agent-in-the-loop workflows
Local-first, no telemetry design
Support for custom enterprise CLI tools via plugin system
91.8% total token savings across 2 months of personal use (4.4M raw → 4.1M saved)
kubectl get: 93.9% savings (101 runs)
grep: 96.2% savings (103 runs)
docker: 96.1% savings (127 runs)
Pluggable, open-source CLI filter with local-first design (no telemetry)
Designed as agent hook / shell wrapper for reducing LLM input size
Supports custom filters for enterprise/internal CLI tools
Go to the source
Hacker Newsgithub.com
Publisher excerpt: Hi HN, Not sure if anyone would be interested. But, just wanted to share that I've been maintaining my small tool called 'lowfat' that helps me filters some of my verbose CLI output. It's a single binary, works as an agent hook or a shell wrapper. It has a plugin system to customize filters per…