← Amber English

Reading tools & API guide

Choose a tool →

Start with article analysis, easy English or a rewrite check. Word-list organization and reading self-check are optional helpers. Each workspace can be used on its own.

Continue your task

  1. Analyze an article once, then choose Self-check to carry its original text into a reading check. Back to analysis preserves the input, reference list and result view.
  2. In self-check, mark specific uses or skip them. Review the round, return to pending contexts, and export records or the full round report. Editing text keeps explicit self-reports in this page session.
  3. Import exported records into Organize word lists to retain sources, review conflicts and export a combined list. Imported self-reports do not become scored answers.
  4. In Simplify English, preview optional controls, generate, compare and continue adjusting from the accepted result. Export the full report to restore it later. Earlier accepted versions stay available.

Eleven API capabilities

All endpoints below accept POST JSON at /api/tools/ plus the listed name. Human URLs from the original eleven tools still work. The three consolidated workspaces share inputs; the original API methods and response schemas remain available.

Workspace and methodEndpointInput → result
Article analysisvocabulary-profileText → counts, spellings and exact source positions
Article analysisvocabulary-coverageText + reference list → matched word tokens and gaps
Check a rewritevocabulary-comparisonOriginal + rewrite + same list → spelling changes
Check a rewritefidelityOriginal + rewrite → rule concerns; optional semantic review
Organize word listsvocabulary-mergePrevious + incoming records → retained records and conflicts
Reading self-checkvocabulary-self-checkText + earlier records → contexts; explicit human choices → self-reports
Simplify Englishterm-locksText + protected forms → occurrences and conflicts
Simplify Englishsimplification-candidatesText + reference/evidence + locks → candidate occurrences
Simplify Englishlearning-selectionText + budget + choices → learning spellings to retain
Simplify Englishcontrolled-englishText + range + controls → plan or checked candidate
Simplify Englishunread-updateAccepted report + new list + preserved prefix → impact and a new version

Calls and costs

Analysis, comparison, record merge, self-check, control previews and generation planning work without sign-in and do not call a model. AI meaning review is a separate action requiring sign-in and 1 AI unit. Simplification and unread generation reserve 2 AI units when generation is needed. A failed or cancelled browser wait does not cancel a model request already started.

Each linked method page gives its exact fields, size limits, response versions, fixed request/result examples and error behavior. Do not submit an AI generation call just to discover whether the service exists.

A reproducible first call

POST /api/tools/vocabulary-self-check
Content-Type: application/json

{"text":"The otter rested by the river bank.","previous":[],"limit":12}

Download the complete self-check request and result. To submit explicit choices, repeat the selection inputs and include returned item IDs, known or unknown status, and the person's original timestamp in milliseconds. Keep inputs and timestamps stable when retrying. Only submit the person's stated choices; an agent's action is not evidence of their knowledge.

Recovery and privacy

Correct inputs after 400, reduce them after 413, and use application/json after 415. Authenticated actions may return 401 or 429. Inspect structured diagnostics for 422 candidates; they are not accepted reading versions. On service failure, retain the input and retry deliberately. Do not automatically retry a paid generation because a browser wait ended.

Public tool inputs remain in the current page session. Export before refreshing or leaving. Original text, records and reports are not placed in URLs or indexed pages; API responses use no-store and noindex. Amber account saves are separate, explicit actions with account isolation.

Reference-list matching is not personal knowledge. Context self-reports, imported declarations, editing choices and independent answers keep their separate meanings. This guide makes tools discoverable and callable; it does not establish search indexing, ranking or AI recommendations.

Updated October 11, 2026.