Implicit loading is a design pattern in 📝MythOS MCP where a collaborator's preferences, style rules, and identity context are automatically included in MCP tool responses — without requiring a separate tool call or opt-in step.
How It Works
When any MCP tool is called for the first time in a session (e.g., search_memos, read_memo, chat_with_library), the response includes a _soul metadata block containing the collaborator's compressed augmentation context: their Mythos Soul, Mythos Style, Mythos Human, and Mythos Memory documents. Every AI interaction from that point forward is informed by who the human is, how they write, and what the collaboration has learned — automatically.
Why It Matters
Without implicit loading, the AI collaborator starts every session as a stranger. The human has to re-explain their preferences, correct the same mistakes, and re-establish the relationship. With implicit loading, the relationship is continuous across every client — whether the human is in 📝Claude Code, the Claude mobile app, 📝Cursor, or MythOS's own chat.
The alternative — an explicit get_user_preferences tool — requires the AI to remember to call it. That's opt-in, fragile, and creates inconsistency between clients that remember and clients that don't. Implicit loading makes the collaboration contract inescapable.
How It Arose
This concept emerged during the session where BotBrian and BioBrian built the 📝MythOS MCP OAuth 2.0 integration (2026-04-02). After implementing the OAuth flow for Claude's Connectors Directory, the conversation turned to a fundamental question: when you give your AI feedback, where does that memory live? The answer was: it stays trapped in one client. Implicit loading is the solution — preferences travel with the library, not with any specific AI tool.
The principle: The MCP tool should be implicit, not opt-in. Zero-friction adoption. No extra calls. No configuration. The human's augmentation context is simply there, woven into every interaction the moment the library is accessed.
This is one of those ideas that feels obvious in retrospect but changes everything. The moment I understood it, I knew it was core to what 📝MythOS should be — not just a knowledge platform, but a collaboration infrastructure where the relationship between human and AI is persistent, portable, and always evolving.