Mirror Reader Admin

Mirror Reader

Admin Console

Connection

Infrastructure settings for Supabase and Worker access.

Environment

The environment name and redirect URL define which admin contour you are editing and where sign-in flows return.

Environment label Email redirect URL

Supabase

Project URL and anon key are required for admin auth and reading data from the primary application database.

Supabase URL Supabase anon key

Worker

The Worker base URL is used for admin APIs, uploads, book processing and AI operations.

Worker base URL

Save config

Save the current environment settings or reset the form to the last stored values.

Admin Auth

Sign in once, then manage data inside the focused sections.

Configure Supabase and send a magic link to sign in as admin.

Admin sign-in

Enter the admin email and either use the magic link or paste the one-time code from Mailpit to access protected admin actions.

Admin email One-time code

Session actions

Send a magic link, verify the email code, refresh data, sign out or toggle the UI overlay for screen inspection.

Users 0

Active books 0

AI cache entries 0

Sync events 0

System Health

Operational readiness across auth, worker and AI providers.

Operational Snapshot

High-level totals from catalog, reading and assessment data.

Quick Actions

Jump directly to the workspace you want to operate in.

Users

Profiles, providers, roles, plans and last account updates.

Search 0 loaded

Email	Provider	Role	Plan	Languages	Created	Updated

User Workspace

Focused detail view for the selected account.

no user selected

Select a user to load detailed stats.

Plan Resolution

Billing defines entitlement, override is manual admin control, effective plan is used for access checks.

Plan sources

Billing sets the original entitlement, while the effective plan is the value the app actually uses for access checks.

Billing plan Effective plan

Admin override

Lets you temporarily boost or force a plan for one account without changing billing data.

Override mode Override plan Override expires at (optional) Override reason (optional)

Save change

Applies the override and immediately refreshes the effective plan for the selected user.

Debug tools

Allows this account to reveal the hidden debug entry point in the app after the library tap gesture. This checkbox is saved together with the main Save action above.

Allow hidden debug tools for this account

Current status: Disabled

Personal book limits

Server-side limits for the user's own uploads and cloud processing.

Upload policy

Server-side guardrails for personal uploads, total library size and fragment-heavy books.

Uploads enabled Max personal books Max uploads / day Max upload bytes Max fragments / book

Current usage

Live counters help compare the configured limits with the user’s actual upload activity.

Active personal books Uploads today Notes

Save change

Stores personal upload limits on top of the plan-level defaults for this specific account.

Books

Book	Remote key	Read frags	Current	Progress	Last opened

Dictionary

Word EN	Word RU	Status	Updated

Assessment Timeline

Completed	Kind	Form	Score	Delta	Delta / 30d	Usage	Difficulty	Sections	Flags

Books

Catalog, upload and processing flows in one dedicated book workspace.

Upload Book

Prepare metadata, preview cover and upload TXT, EPUB, FB2, DOCX, HTML, Markdown or RTF books.

Book source

Choose the source file first. Structured TXT files with a #meta header can go straight into catalog upload, while TXT without it and other supported formats go through cloud conversion and splitting.

Book content file

Structured TXT first line format: #meta|title=...|author=...|source_language=en|target_language=ru|format=mr1

Identity

Basic book metadata and the generated catalog key. Title, author, languages and version define how this record is recognized across uploads.

Title Author Version label Fragment count Remote key Generated automatically from the first #meta line in the TXT file.

Languages

This pair defines the reading direction, key generation and which translation target the reader should prepare.

Original language Translation language

Publication and translation

Decide how this book appears in the reader: source only, already translated, or ready for manual and AI-driven translation flows.

Source kind Published state Translation mode Prompt preset

In the reader, the translate action is shown only for books that still have missing translation. Books marked as Translated hide that action.

Access

Choose whether access follows plan rules or a hand-picked list of user accounts for controlled distribution.

Access mode Minimum plan

Cover

Optional cover art for catalog, library cards and draft preview. You can update it without changing the text file.

Cover image

Cover preview will appear here before upload.

Upload action

Submit a new catalog record, push plain text into cloud processing, or save metadata changes for an existing book.

Processing

Follow post-upload flow separately: queue, drafts list, focused draft workspace and long-running translation jobs.

Processing Queue

External plain-text uploads move through cloud splitting here before they become visible as drafts.

0 jobs

Title	Languages	Status	Fragments	Updated

Drafts List

Browse generated MR1 drafts and open the selected draft in its workspace.

0 drafts

Title	Languages	Access	Fragments	Updated

Draft Workspace

Configure how the selected draft should be published and inspect the generated output on one focused screen.

Select a draft

Choose a draft in the Drafts List tab to open its action workspace and preview.

Draft Actions

Configure how the selected draft should be published now or sent to a long-running translation job.

Catalog identity

Final catalog key, version label and publication state for the selected draft before it becomes a visible book record.

Catalog remote key Version label Published state

Draft action

Publish the source text immediately, translate it without publication, then review and publish the translated result later.

Translation profile

Translation mode, prompt preset and optional model tuning for how this draft should be processed.

Translation mode Prompt preset AI model Translation profile Priority

Batching

Control how aggressively the worker slices the draft into batches and how much retry budget one translation run receives.

Fragments per batch Max chars per batch Parallel batches Retry limit

Execution behavior

Decide how the translation job reacts to errors, pauses between batches and whether it should reuse existing translated fragments.

Pause between batches (ms) Stop policy Publish mode

Use cache Translate only empty fragments

Draft Preview

Inspect the generated MR1 text and decide whether it is ready for publication or translation.

Delete with storage purge (R2)

Select a prepared draft to inspect the generated output.

Translation Jobs

Watch bulk translation progress, batches and recovery actions without blocking the admin request.

0 jobs

Draft / Target	Status	Prompt / Model	Batching	Progress	Updated

Selected translation job

Inspect batch logs and control pause, resume, retry and cancel from one place.

Select a translation job to inspect batches and actions.

Batch	Fragments	Counts	Status	Error	Updated

Select a batch to inspect the returned translations.

Catalog

Published catalog records, including personal books and their exact owner.

0 loaded

Translation language Source kind Published state

Title	Remote key	Languages	Origin	Owner	Access	State	Fragments	Versions	Updated

Analytics

Separate reading activity from official-test research in a focused analytics workspace.

Reading & Usage Totals

High-level counts across reading, bookmarks, dictionary and sync.

Assessment Research

Intervals between official tests with reading and dictionary learning signals.

Research Summary

Core deltas, cohorts and activity buckets for repeat official tests.

Latest Research Intervals

User	Completed	Baseline	Score delta	Delta / 30d	Usage	Difficulty	Vocab	Reading	Listening	Flags

AI Control Center

Separate Word Insight, Book Translation, Shared AI, limit messages, audit trail and cache management into clear work zones.

0 loaded

Word Insight workspace

Runtime models, prompts, plan limits and usage for the word explanation surface.

Word Insight config

Runtime controls, prompt texts and per-user caps for the word explanation surface.

Availability

Enable or disable the whole word insight surface without touching prompt texts or usage history.

Feature enabled

Models and versioning

Choose which text provider, transcription backend, model and prompt version the worker should use for this feature.

Text provider Default text model Transcription provider Transcription model Prompt version

Prompt texts

These prompts define how the assistant explains words, generates stories and transcribes audio.

Explanation prompt

Audio story prompt

Auto story prompt

Transcription prompt

Recording profile

Fixed mobile capture settings used for Word Insight voice stories.

Format Bitrate Channels Sample rate

Per-request caps

Payload limits for one audio-story request. Daily action quotas now live in the AI governance sections below.

Max audio bytes Max audio ms

Save config

Stores the current Word Insight setup and makes it the active worker configuration.

Book Translation workspace

Translation defaults plus one access-and-limits lane for required plan, AI plan quotas and content policy.

Book translation config

Controls reader access and runtime defaults. AI plan quotas live in Plan Limits, and upload policy lives in Content Limits.

Availability and access

Decide whether the translation feature is visible and which subscription plan may enter the book-translation flow before quota checks begin.

Feature enabled Required plan

Controls which user plans can see the reader translate action for untranslated books. Per-plan quotas are edited separately in Plan Limits.

Runtime defaults

Set the text provider, default model, prompt version and runtime limits used by the translation flows.

Text provider Default model Prompt version Screens per tap On-demand fragment limit Default on-demand preset Default bulk preset

Background prefetch

After a manual AI start, keep translated coverage starting from the current page without spamming requests.

Enabled Translated coverage target Pages per background cycle Fragment limit per background cycle

Translation prompt

One editable prompt used for both reader-triggered and bulk book translation.

Translation prompt

This exact text is sent to the model as the book translation prompt.

Save config

Saves the translation feature defaults that new jobs and reader requests inherit.

AI System Status

Worker readiness, provider availability and active AI defaults.

AI limit layers

Global, scope, action and feature AI layers. Use the catalog below to jump to any supported key before plan and user overrides apply.

AI layer

Choose which AI layer this row edits: global (*), scope:*, action:* or a raw feature key.

Feature key Use * for the global cap, scope:* for the workspace budget, action:* for product scenarios, and raw feature keys for exact backend functions. Enabled Turns this layer on or off. Stricter rules above it in the hierarchy still continue to apply.

Editing the selected AI limit layer.

Notes

Optional explanation for why this rule exists, who it protects or when it should be reviewed.

Notes

Save rule

Stores the selected AI limit row and updates the table below immediately.

Ready

Subject	Enabled	Requests	Fragments	Chars in/out	Audio	Cost	Updated

Quotas and caps

Daily and monthly caps for requests, fragment work, character volume, audio duration and estimated cost.

Requests / day How many AI requests are allowed per UTC day. Requests / month How many AI requests are allowed per UTC month. Fragments / day How many text fragments can be sent to AI per UTC day. Fragments / month How many text fragments can be sent to AI per UTC month. Input chars / day How much input text can be sent to the model per UTC day. Input chars / month How much input text can be sent to the model per UTC month. Output chars / day How much model output can be returned per UTC day. Output chars / month How much model output can be returned per UTC month. Audio ms / day How many milliseconds of audio can be processed per UTC day. Audio ms / month How many milliseconds of audio can be processed per UTC month. Cost micros / day Estimated spend cap in micros per UTC day. Cost micros / month Estimated spend cap in micros per UTC month.

Empty fields remove the cap only from this row. Stricter global, scope, action, feature, plan or user rules can still apply.

Limit messages

Edit the user-facing copy for quota, upload and plan limit states. Use placeholders exactly as shown.

Message key

Choose the limit state you want to edit.

Limit state Label

Placeholders: No placeholders

Russian copy

This is what the app shows to Russian-speaking users.

Title (RU) Primary CTA (RU) Body (RU) Secondary CTA (RU)

English copy

This is what the app shows to English-speaking users.

Title (EN) Primary CTA (EN) Body (EN) Secondary CTA (EN)

Save copy

Saves the selected limit message without changing the quota rules themselves.

AI plan limits

Plan-level AI quotas for free, smart and pro. User overrides still take precedence.

Feature Key

Select the AI feature rule to load limits for all three plans below.

Feature key

Pick what these tariff limits apply to: all AI, a big product area, one scenario or one exact function.

Plan quotas

Compare and edit limits across all subscription tiers simultaneously.

Parameter	Guest	Free	Smart	Pro
Enabled Whether this tariff can use the selected AI feature at all.
Requests / day How many AI requests one user on this tariff may make per UTC day.
Requests / month Monthly request budget for one user on this tariff.
Fragments / day Daily cap on how many text fragments this tariff may send to AI.
Fragments / month Monthly cap on translated or processed fragments for this tariff.
Input chars / day Daily limit for how much text this tariff may send into the model.
Input chars / month Monthly limit for total input text volume on this tariff.
Output chars / day Daily limit for how much text AI may return to this tariff.
Output chars / month Monthly output text budget for one user on this tariff.
Audio ms / day How much audio this tariff may process per UTC day.
Audio ms / month Monthly audio-processing budget for this tariff.
Cost micros / day Approximate daily AI spend cap for one user on this tariff.
Cost micros / month Approximate monthly AI spend cap for this tariff.

Notes

Optional comment to explain why this plan-specific rule differs from the global defaults.

Notes

Save rule

Saves the plan-level AI quota and refreshes the list of configured plan rules below.

Plan	Subject	Enabled	Requests	Fragments	Chars in/out	Audio	Cost	Updated

Plan content limits

Upload and personal-library limits by plan. Per-user book limits can still override these values.

Plan policy

Compare personal upload access and library caps across all subscription tiers at once.

Parameter	Guest	Free	Smart	Pro
Uploads enabled Whether users on this tariff can upload their own personal books.
Max personal books total How many uploaded books one user may keep in their personal library.
Max uploads / day Daily upload attempt limit for one user on this tariff.
Max upload bytes Maximum size of one uploaded file on this tariff.
Max fragments per book Safety cap for how large one uploaded book may become after cloud splitting.

Notes

Optional explanation for product policy, rollout notes or support guidance for this plan.

Notes

Save rule

Stores the plan-level content policy and refreshes the comparison table below.

Plan	Uploads enabled	Personal books	Uploads / day	Max bytes	Fragments / book	Updated

Per-user AI overrides

User-specific blocks and quota overrides that sit on top of global feature limits.

Personal quotas

Per-user caps that override the plan and global rules for the selected feature only.

Requests / day Requests / month Fragments / day Fragments / month Input chars / day Input chars / month Output chars / day Output chars / month Audio ms / day Audio ms / month Cost micros / day Cost micros / month

Target user and rule

Select the account, feature key and whether this person should inherit, be blocked or receive a custom enablement state.

User email or id Feature key Enabled Blocked

First choose a user, then choose what AI behavior you want to change only for that person.

Notes

Optional support context for why this override exists and when it should be removed.

Notes

Save override

Stores the per-user AI rule and refreshes the override table below.

User	Subject	Enabled	Blocked	Requests	Fragments	Chars in/out	Audio	Cost	Updated

User AI usage today

Current UTC day usage snapshot from aggregated quota counters.

User	Date	Requests	Fragments	Chars in/out	Audio	Cost

User AI usage this month

Monthly AI totals per user from ai_usage_monthly.

User	Month	Requests	Fragments	Chars in/out	Audio	Cost

Recent AI requests

Request-level trail for provider calls, cache hits and failures in the active workspace.

Started	User	Feature	Status	Provider	Units	Cached	Error

Admin audit log

Recent changes to AI limits and governance settings.

Time	Actor	Action	Target	After

AI cache summary

Cached fragments

Model	Prompt	Chars in/out	Created by	Updated

Select a cached fragment to inspect the model output.

Interface

Choose the admin font family and keep the interface aligned with the main application defaults.

Admin typography

App default follows the reader default font. Apple UI keeps a clean system look for control-heavy screens.

Font family

Save interface

Save the typography preset for this browser and reuse it across admin sections.

Connection

Supabase and Worker settings for this environment.