Transparent calculations
Counts, percentages, and per-1,000 rates stay visible so you can verify what each result means.
WEB · API · CLI · MCP
For Humans and AI Agents
Use transparent text analysis in the browser, automate it through a stateless API, or run the same deterministic methods locally from a terminal or AI agent.
Pasted text is processed without server storage.Load an example, paste your text, and inspect a transparent result without writing code.
Browse the eight tools→DEVELOPERSUse stable JSON endpoints or the local-first npm CLI in scripts, CI, and editorial workflows.
Read developer docs→AI AGENTSWorks with MCP-compatible agents such as Codex, Claude Code, and Gemini CLI. They can call precise, read-only analysis tools instead of guessing word counts.
Open agent integrations→AVAILABLE NOW
Each tool has a focused workflow, visible formulas, and a result you can inspect instead of a single opaque score.
Count every word, search and sort the full vocabulary, edit stop words, and export CSV or JSON.
Open tool→Measure words, bigrams, and trigrams, track exact phrases, and compare density changes between two results.
Open tool→Review vocabulary, bigrams, tracked phrases, and Zipf distribution, then compare result A with result B.
Open tool→Compare two texts or webpages by word count, vocabulary, normalized word frequency, bigrams, and Zipf diagnostics.
Open tool→Analyze recurring phrases of custom length (including bigrams and trigrams) in text or a public webpage.
Open tool→Build raw term vectors from text or a URL and inspect term frequencies, percentages, and export-ready rows.
Open tool→Calculate corpus-aware TF-IDF scores for 2–10 documents and compare weighted term influence.
Open tool→Measure cosine similarity between two texts with BoW or TF-IDF, then inspect top contribution terms.
Open tool→WHY THIS TOOLKIT
Counts, percentages, and per-1,000 rates stay visible so you can verify what each result means.
Text-only work stays in your browser. If a workflow includes a public URL, required inputs use the stateless API and are not stored.
Normalize texts of different lengths and compare drafts, pages, or versions using the same settings.
The tools expose patterns and repetition without pretending that one percentage determines content quality.
LEARN THE METHODS
Counts, percentages, per-1,000 rates, tokenization, and a worked example.
GUIDEExact phrase calculations, comparison rules, and why density is not a ranking score.
NLP GUIDEHow frequency tables become document vectors and machine-learning features.
NLP GUIDETerm frequency, smoothed inverse document frequency, corpus effects, and a worked example.
NLP GUIDEHow document vectors become a 0–1 overlap score and what that score cannot prove.
NLP GUIDEUnigrams through longer phrase windows, denominators, filtering, and practical uses.
WORKFLOW GUIDEMeasure normalized vocabulary changes without confusing them with a character diff.
ONE ENGINE · MULTIPLE INTERFACES
Call the stateless JSON API, run eight local-first npm commands, or expose the same read-only operations through MCP. Every interface uses the same transparent analysis methods.