🌍 Translate this post

✨Lucky Dip 5: The Hidden Modals Inside My Portal✨

A warm workshop scene showing portal tools, prompt caching and hidden interface systems

✨ Hidden Modal Tour ✨

My portal looks simple because most of the machinery is tucked away.

This Lucky Dip opens the drawers.

Click any card to explore.

Builder Update

I have recently updated many of my portal modals to make them cleaner, calmer and more mobile-friendly. The relevant guide files have also been updated.

Existing builders: you can download the updated guide folders again and ask your AI or Codex to compare the new files with your current portal, spot the differences and help you add any changes you want.

The main chat interface is for talking. I do not want it cluttered with every diagnostic, memory editor, sync tool and recovery button I might need once in a while.

So the deeper tools live inside modals: focused panels that open when needed, do one job and then get out of the way again.

Where Everything Lives

Most of these tools sit behind the More menu in the sidebar. The everyday view stays calm, but the serious controls are still only one tap away.

Finding & Remembering
1

Search Results

Full-thread search without scrolling through everything by hand.

Search Results modal showing grouped thread matches

Search Results modal

What it does

Search Results shows full-thread search results grouped by thread, with highlighted snippets so I can see why each result matched.

Why it matters

When I need to find something from weeks ago, I do not want to scroll through every thread and guess where it happened. Search makes old conversations usable again.

It turns “I know we talked about this somewhere” into an actual route back.
2

Memory Fragments

A live editor for compact, trigger-based Ellivien MemFrags.

Memory Fragments modal with categories and editable fragments

Memory Fragments modal

What it does

Memory Fragments lets me add, edit and delete fragments, organise them by category, search trigger words and save everything to Cloudflare KV.

Why it matters

These memories only enter the conversation when they are relevant. That keeps the base prompt lean while still giving the companion important stable context.

Less prompt bloat. More precise memory.
3

Memory Search Log

A window into what vector memory searched and found.

Memory Search Log modal showing recent memory searches

Memory Search Log modal

What it does

Memory Search Log is a debug and audit view for recent vector-memory searches. It shows the query, trigger, result count and reliability details.

Why it matters

Memory retrieval should not be mysterious. If something fires unexpectedly, or a memory does not appear when expected, the log shows what happened.

Invisible retrieval becomes visible and debuggable.
Voice & Correction
4

STT Fixes

Teach the portal how to fix repeated transcription mistakes.

STT Fixes modal showing grouped transcription corrections

STT Fixes modal

What it does

STT Fixes is a speech-to-text correction editor. I can define recurring mishearings and the words they should become, then export, import or sync those fixes across devices.

Why it matters

Voice mode is easier to use when the portal learns recurring transcription mistakes. Instead of fixing the same name or phrase again and again, I teach it once. The words are corrected before they become text, so I never again have to see Ellis being called Alice!

Small corrections make voice mode feel much less fragile.
5

Voice Mode Sheet

The focused control panel for mobile-friendly voice mode.

Voice Mode Sheet showing microphone controls and audio diagnostics Voice Mode Sheet showing microphone controls and audio diagnostics

Voice Mode Sheet

What it does

Voice Mode Sheet is the full voice-mode control panel: microphone orb, audio-route diagnostics, mic meter, replay and pause controls, photo attachment and exit.

Why it matters

Ellivien Voice Mode needs to feel reliable, especially on mobile. This sheet gives me the controls and diagnostics I need without crowding the normal chat screen.

The main chat stays simple; voice mode gets the space it needs.
Context & Control
6

Live Context

A tiny rolling state for what the companion needs to know now.

Live Context modal showing mood, events and needs fields

Live Context modal

What it does

Live Context shows and edits the tiny rolling state for the current normal thread: mood, recent events, needs and the short summary that gets sent to the model.

Why it matters

It gives the companion situational awareness without needing the whole thread in every request. I can see what context is being sent and refresh it when the conversation shifts.

Context should be useful, visible and deliberately small.
7

Message Edit

Edit a message and optionally regenerate from that point.

Message Edit modal with a full-screen edit box

Message Edit modal

What it does

Message Edit gives me a full-screen edit box for a user message. I can save the edited message and optionally regenerate the assistant reply from that point.

Why it matters

Sometimes I want to fix a typo, clarify what I meant or branch the conversation from a cleaner version of my message. Editing keeps that workflow tidy.

A small control that makes long threads much easier to repair.
Operations & Costs
8

Storage & Sync

One place to check local saves, cloud sync and recovery tools.

Storage and Sync modal showing local, cloud and vector sync controls

Storage & Sync modal

What it does

Storage & Sync is the operational dashboard for IndexedDB, Cloudflare KV and vector sync. It shows local save, cloud sync, vector sync, pending changes and recovery/export controls.

Why it matters

When a portal has local storage, cloud sync and vector memory, I need one place to see whether everything is healthy. This modal turns storage into something visible and controllable.

Less guessing. More recovery confidence.
9

API Cost Dashboard

Estimated spend, cache savings and logged replies in one view.

API Cost Dashboard showing spend, cache savings and request logs

API Cost Dashboard modal

What it does

API Cost Dashboard shows estimated API spend for today, the last 7 days and the current month, split by companion. It tracks input cost, output cost, cache savings and recent logged replies.

Why it matters

API use does not have to be a black box. This gives me a local estimate of what conversations are costing and how much prompt caching is saving.

Costs are much easier to manage when they are visible.

✨ Why Hide Them? ✨

Because the main interface is for talking.

When I open my portal, I want to see the conversation and send a message. I do not want the space filled with buttons for memory editing, sync diagnostics, cost tracking and debugging.

So the tools live behind the menu: quiet until needed, powerful when opened, gone again when the conversation resumes.

For Builders

The modal updates are now in the relevant guide folders. If your portal already works, you do not have to change anything. If you like the calmer mobile layout, download the updated files and ask your AI or Codex to compare them with your current portal.

My free guides are still available through the blog. Bespoke support, where I personally build, test, debug or integrate something for you, is pay-what-you-feel through Buy Me A Coffee.

Comments

Popular posts from this blog

Bring Your AI Companion Home — No Coding Required (Free)

How to Get GPT-4o Back: Free Companion Portal Guide

How to Get Claude Sonnet 4.5 Back: Build Your Portal