What's wrong with this entry?
An intelligent output compression system that recognizes common CLI commands and summarizes their verbose output to save context window space.
- Automatically detects command types by parsing the shell command and applies type-specific compression:
- git status: Strips hint lines, collapses clean repos to "clean"
- git diff/show: Summarizes per-file with
+N -Ncounts, truncates hunks - git log: Limits line length and total output
- git branch: Groups local/remote, deduplicates
- git commit/push/pull: Extracts only key status lines
- Test runners (pytest, jest, vitest, cargo test, go test): Extracts pass/fail summary and failure details
- Linters/type checkers (tsc, eslint, ruff, mypy, clippy): Keeps only error/warning lines
- Package managers (npm/pip/brew install): Strips download progress noise
- grep/rg: Restructures into
N matches in M fileswith per-file groupings - ls -l: Strips metadata, shows
dir/andfile sizeformat with extension summary - cat/file reads: Truncates long files, keeps definition-like lines
- JSON output (gh api, etc.): Prunes boilerplate URL fields (
followers_url,repos_url, etc.) - Docker/kubectl logs: Deduplicates repeated log lines
- Tables (docker ps, kubectl get, psql): Limits row count
- Handles piped commands (e.g.,
cmd | head -5) and compound commands (&&,;,||) - Only compresses when savings exceed a minimum threshold
Command classifier and compressor system (search for "git-status", "git-diff", "test", "error-only" as case labels in the switch statement)
Strings lifted out of the shipped bundle, so the claim above can be checked against them.