tokensaver scan flags token waste across every prompt file in a repo
The tokensaver scan tool identifies and flags token waste across prompt files in a repository, helping developers manage costs associated with growing prompt sizes. It measures token usage for each prompt file, suggesting fixes and displaying potential savings. The tool supports 20 models, providing exact counts for OpenAI and estimations for Claude and Gemini. It can be integrated into CI/CD pipelines using the --json option to fail jobs when prompts exceed a budget, and is installed via pip install prompt-tokensaver.
Time & source
Times shown in UTC
Display time zone: UTC
Local time zone unavailable; showing UTC.
PublishedOffset at this time: UTC+0Oct 30, 2026, 00:15 UTC
IngestedOffset at this time: UTC+0Oct 30, 2026, 02:00 UTC
- Published
- Oct 30, 2026, 00:15
- Ingested
- Oct 30, 2026, 02:00
- Source type
- Dev community
- Tier
- Community
- Source status
- Healthy
Tier is a per-source editorial setting, not a per-item score.
Our prompts live in the repo as text files, and they kept growing: a copied rule here, a pretty-printed schema there. Nobody noticed until the bill came.
tokensaver scan prompts/ measures every prompt file and shows what the safe fixes and all fixes would save per file, biggest saving first, with a total at the bottom. Add --json and you can fail a CI job when a prompt goes over budget. It's offline and covers 20 models. Counts are exact for OpenAI and estimated for Claude and Gemini.
pip install prompt-tokensaver · https://github.com/behrouzbk/tokensaver