From token-saver
Configures and diagnoses token-saver compression settings including compression levels, processor status, hook debugging, and savings statistics.
How this skill is triggered — by the user, by Claude, or both
Slash command
/token-saver:token-saver-configThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Run `token-saver stats` to see compression statistics for the current and all sessions.
Run token-saver stats to see compression statistics for the current and all sessions.
Token-saver config is stored in ~/.token-saver/config.json. Available settings:
min_lines: minimum output lines to trigger compression (default: 5)min_chars: minimum output chars to trigger compression (default: 200)chars_per_token: ratio for token estimation (default: 3.5)wrap_timeout: max seconds for command execution (default: 30)To modify: token-saver config set min_lines 10
Set TOKEN_SAVER_DEBUG=true environment variable to enable debug logging to ~/.token-saver/hook.log.
Token-saver includes processors for: git, gh (GitHub CLI), docker, kubectl, terraform, npm/pip/cargo, test runners (pytest, jest, go test), linters (eslint, ruff, pylint), build tools, cloud CLIs (aws, gcloud, az), database queries, file listings, file content, environment/system info, network tools (curl, wget), and search (grep, find, ripgrep).
If compression isn't working:
python3 is available in your PATHTOKEN_SAVER_DEBUG=true then trigger a compressible command~/.token-saver/hook.log for errorsecho "test" | python3 "${CLAUDE_PLUGIN_ROOT}/scripts/hook_pretool.py"npx claudepluginhub ppgranger/token-saver --plugin token-saverMonitors Headroom proxy health, reports compression stats (30-40% token savings), and troubleshoots connection issues for Claude Code sessions.
Minimizes token waste in bash, file reads, and data processing using jq for JSON, yq for YAML/TOML, awk for CSV, and ast-grep for precise searches.
Applies token optimization rules to reduce context usage and response length. Always active, it minimizes file reads, avoids preamble, and batches tool calls.