tokensift
A local static-analysis linter that finds wasted tokens in LLM prompts and payloads.
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
tokensift is a token-efficiency linter for LLM prompts and payloads. It uses deterministic local static analysis to inspect prompt strings, Message arrays, and tool schemas. It can be used as a library or CLI and integrated with CI.
KEY FEATURES
Key features
- Tokenizer-level analysis of prompts, message arrays, and tool schemas
- Shows where tokens are wasted and the actual cost
- Library and CLI support
- Prompt file checks in CI
EDITOR'S NOTE
Why did it stand out?
Using actual tokenizers rather than character counts, it identifies wasted tokens and their cost, making prompt optimization a linting task. It distinguishes exact provider-specific calculations from estimates to convey the certainty of its results.