Components › Context Management › Active Memory Panel
CM03 · Context

Active Memory Panel

Purpose Displays the AI's current working memory, references, or persistent knowledge being used in this session.
Use when The AI system has session memory, user preferences, prior conversation references, or persistent facts that influence its behavior beyond what the user explicitly typed in this turn.
Do not use To expose model weights or training data — that is not memory, it is architecture. Do not use when memory state is opaque, inaccessible, or cannot be reliably read by the system. If you cannot accurately surface what the AI is drawing from, rendering this component gives users false confidence — they believe they understand what the AI knows when they do not. Inaccurate memory disclosure is worse than no disclosure.
Failure mode prevented Users receive responses influenced by prior sessions, stored preferences, or injected context they are not aware of. Unexplained behavioral differences undermine trust and make debugging impossible. The inverse failure is equally serious: a memory panel that is incomplete, stale, or inaccurate actively misleads users into thinking they have oversight when they do not.
Example Dark Academia
Dark Academia
The Atelier
Neon Tide
The Archive
Candlelight
The Conservatory
Memory Active session memory 5 entries
  • 📌
    User role Platform engineer, SnapLogic integrations team
    Profile
  • 📌
    Preferred output format Bullet summary followed by detailed breakdown
    Learned · 3 sessions ago
  • 📌
    Active project Q1 pipeline reliability review
    This session
  • 📌
    Last referenced file incident-template.md — partially indexed
    2 min ago
  • 📌
    Prior session context Last session analyzed Salesforce connector failures; follow-up on Workday requested
    Yesterday
Overview
memory-panel-full — complete list of memory entries with timestamps and sourcesmemory-panel-compact — count + summary with expandable detailmemory-panel-editable — allows users to delete or modify memory entries
Design rule

Every memory entry must show its source — where the AI got this information. "Learned" without a timestamp, "From profile" without a reference, or "Remembered" without recency is insufficient. Users must be able to trace each memory to its origin.

Theming API

Override these CSS custom properties to apply your own visual treatment without touching the component's behavioral contracts.

Shared semantic tokens

Token Controls
--rad-surface-card Component card background
--rad-border-subtle Card border
--rad-color-text-primary Panel title and key labels
--rad-color-text-secondary Memory value text
--rad-color-text-meta Source timestamp
--rad-color-approve Focus ring
--rad-radius-card Card border radius

Component-scoped tokens

Token Controls
--rad-memory-entry-border Row separator
--rad-memory-key-color Memory key label
--rad-memory-value-color Memory value text
--rad-memory-source-color Source provenance label
--rad-memory-delete-color Remove button default
--rad-memory-delete-hover Remove button hover
--rad-memory-count-color Entry count label

"*If the AI remembers something about you, you should be able to see what it remembers — and forget it on your terms.*"

← Previous Context Pills Next → Context Scope Selector

Intellectual Property

RAD is the original work of Jackie Curry. All rights reserved. No portion may be reproduced, adapted, or incorporated into any product or system without express written permission.

Permitted: citation in academic or editorial contexts with full attribution.

Licensing

© 2025 Jackie Curry. All rights reserved. Publication date: 2025.

For licensing inquiries, connect on LinkedIn →