Brain System Transparency

Most AI assistants are black boxes. You type something, a response appears, and you have no idea why it said what it said or what informed its thinking. Meggy takes a different approach — it opens the hood and lets you see everything.

The Brain System is Meggy's cognitive core: it holds the system prompt, memory context, loaded skills, vault summaries, and adaptive suggestions that shape every single response. And all of it is visible, editable, and explorable through a set of transparency tools.

The Prompt Playground

Every response Meggy generates starts with a system prompt — a large, structured prompt that tells the AI who it is, what it knows, and how to behave. The Prompt Playground lets you read, edit, and experiment with this prompt in real time.

You can see exactly what goes into a prompt:

• Identity and persona — who is Meggy for this conversation?
• Memory context — what facts and preferences have been loaded?
• Skill instructions — which skills are active and what do they inject?
• Vault summaries — what documents are in context?
• Tool definitions — which tools are available?

Any change you make in the playground takes effect on the next message. This isn't a debug view — it's a live editor.

Memory Mood Ring

The Memory Mood Ring is a visual indicator of how much memory context is loaded for the current conversation. It shows at a glance:

• How full the context window is — are you close to the limit?
• What's taking up space — memory facts, vault documents, skill prompts, or tool definitions?
• Quality of recall — are the loaded memories relevant to the current topic?

Think of it like a fuel gauge for context. When the ring is nearly full, you know the AI has a lot of context to work with — but also that you're approaching the model's token limit and older context may start getting trimmed.

Archaeology View

Every conversation has a history — not just the messages you see, but the full trace of what the AI considered, which tools it called, what results it got, and how it composed its response. The Archaeology View lets you explore this trace.

For any message in a conversation, you can expand its archaeology to see:

• Tool calls — which tools were invoked, what arguments were passed, and what results came back
• Model selection — which LLM processed this request and why
• Token usage — input and output tokens, cost estimate, and cache hits
• Reasoning traces — for models that support extended thinking, the full chain of thought

This is especially valuable when debugging agent behavior or understanding why the AI made a particular decision.

Adaptive Suggestions

Meggy learns from your patterns and surfaces adaptive suggestions — contextual shortcuts based on what you frequently ask or do. These appear as quick-action buttons above the input field.

The suggestion engine considers:

• Time of day — morning routines, evening wind-downs
• Recent conversations — follow-up questions, continuations
• Frequently used tools — your most common workflows
• Active skills — skill-specific actions you've used before

Suggestions adapt over time. The more you use Meggy, the more personalized they become.

Configuration Editor

The Config Editor gives you direct access to Meggy's operational settings from within the brain view:

• Model preferences — which models to use for different tasks
• Memory policies — how aggressively to store and recall facts
• Skill activation — which skills are loaded into the prompt
• Safety settings — content filters and output guardrails

Changes made in the config editor persist across conversations and sync with the main Settings panel.

Why Transparency Matters

AI assistants that hide their reasoning create distrust. When you can see exactly what went into a response — which memories were recalled, which tools were called, which model was used — you can make informed decisions about whether to trust the output.

Meggy's transparency tools aren't just for power users. They're for anyone who wants to understand their AI, tune it to their needs, and build confidence that the assistant is working the way they expect.