context-cost

Local CLI for auditing AI agent skills, conflicts, duplicates, safety risks, and context cost across coding agents.

Measure how many tokens an AI agent pays to read your repo, and verify that cuts actually save tokens.

[ADOPT] CLI proxy that compresses noisy dev-command output (find/ls/grep/git status) before an agent reads it. Single Rust binary, zero deps. Apache-2.0. NOTE: the vendor's '60-90% token savings' is an ADVERTISED figure and does NOT survive independent measurement - see why_now.