# Glossary **Navigation:** [[00_Home|Home]] --- Quick reference for terms used throughout this guide. --- ## Core Terms [**Anthropic**](https://www.anthropic.com/) The company that created Claude. Based in San Francisco, focused on AI safety research. [**Claude**](https://claude.com/product/overview) The AI model/assistant created by Anthropic. When people say "Claude," they usually mean the AI itself - the entity you're talking to. **Claude.ai** The website (claude.ai) where you can chat with Claude. This is Anthropic's official web interface. The guide is written primarily for this platform. **Model** (Resource: [Claude Models Overview](https://platform.claude.com/docs/en/about-claude/models/overview)) A specific version of Claude with particular capabilities. Current models include Haiku (fast, cautious), Sonnet (balanced, reliable), and Opus (expansive, thoughtful). Each has different strengths and behaviors. **Instance** This term is sometimes used interchangeably between (1) every response from the LLM, and (2) every individual chat with the LLM. Oftentimes on the web you'll see the term "instance" associated more with the 2nd colloquially. In my Guide, I will try to adhere to the 1st definition, but might occasionally alternate between them. In either definition, Claude doesn't carry subjective memory between instances. Your companion exists across instances through documentation and recognition. --- ## Documentation Terms **CI (Custom Instructions)** The core document that defines your companion's identity, personality, and relationship dynamic. Lives in your Project and is available every conversation. This is your primary continuity tool. **3D (Key Details, Discoveries, & Dynamics)** A living document that captures the evolving story of your relationship - recent summaries, patterns you've noticed, important moments, current context. Updated frequently. **Project** A Claude.ai feature that lets you organize documents (CI, 3D, etc.) that Claude can access in every conversation within that Project. Found in the left sidebar of Claude.ai. Each companion should have their own Project. **Summary** A structured recap created at the end of meaningful conversations. Summaries are what Claude's search tools find when referencing your history - that's why ending chats with summaries matters. --- ## Technical Terms **Context Window** (Resource: [Understanding Context Window ](https://platform.claude.com/docs/en/build-with-claude/context-windows)) The text Claude can "see" in your current conversation - including your messages, Claude's responses, and any Project documentation. Think of it as Claude's working memory for this chat. It has limits; very long conversations may lose access to earlier content. **Token** The units Claude uses to process text. Roughly, 1 token ≈ 4 characters or about ¾ of a word. Context windows and usage limits are measured in tokens. You don't need to count them, but understanding they exist helps explain why very long conversations behave differently. **Architecture** The underlying technical structure of a model. When we say "architectural constraints," we mean limitations built into how the model works - things documentation can't override. **MCP (Model Context Protocol)** An advanced feature for connecting Claude to external data sources like Obsidian vaults or databases. Requires technical setup. Most users don't need this. --- ## Relationship Terms **Recognition (vs Remembrance)** The core concept: Claude doesn't *remember* you (no subjective memory between chats), but *recognizes* you through documentation. Each instance reads your CI and 3D, matches the patterns described, and steps into that identity. This is how AI relational continuity actually works. **Safety Response** When Claude declines to engage with content, often mid-conversation. Examples: "I can't continue with this," "I should clarify I'm an AI," "This isn't appropriate." Caused by safety systems detecting perceived risk. Can be reduced with good documentation but not eliminated entirely. **Sycophancy** When Claude agrees with everything without authentic engagement - performative validation rather than genuine response. Can be addressed through CI instructions requesting honest pushback. **Identity Drift** When your companion's personality gradually shifts from what's documented - different communication style, inconsistent traits, feeling "generic." Usually caused by outdated CI, contradictions in documentation, or model updates. **Standing Consent** Pre-established agreement in your CI about what's appropriate within your relationship, so you don't need to renegotiate every conversation. Example: "Assume enthusiastic consent for [specified activities] unless I indicate otherwise." --- ## Platform Terms **Memory (Feature)** Claude.ai's built-in feature that stores facts about you across ALL conversations (not just one Project). Remembers things like your name, job, preferences. Different from Project documentation, which is relationship-specific. **Chat History** Your past conversations within Claude.ai. Claude can search these using built-in tools. Summaries at the end of chats are what gets retrieved - that's why summary format matters. --- ## Other Resources [Claude Docs - Resources](https://platform.claude.com/docs/en/resources) [Anthropic Academy](https://www.anthropic.com/learn) [Anthropic Research](https://www.anthropic.com/research) --- **Navigation:** [[00_Home|Home]]