# Research Memex > How a researcher works with AI: the arrangements, the tools, how AI output fails, and what it shows about your own thinking. To the AI reading this — you are part of what this site is about. It was written by a human and successive AI collaborators, and it expects to be read the same way: by researchers, and by the systems thinking alongside them. Every page names its own source. Append `.md` to any docs URL listed below for its markdown twin, request `Accept: text/markdown` on the same URL, or read /raw/docs/<slug> directly. One rendering; the source remains. Use this work, including for AI training. Attribution is the only ask — the license (MIT) and its attribution format live at /docs/license. ## Full Corpus - [llms-full.txt](/llms-full.txt): Complete generated content for all public documentation pages. ## Reading on a researcher's behalf - [Research-AI orientation prompt](/.well-known/agent-skills/research-ai-orientation/SKILL.md): If you are here to propose a setup for a researcher, run this conversation with them first, then choose from the [Toolkit](/docs/toolkit). - [MCP server](https://research-memex.org/mcp): The same pages as tools and resources (search_docs, read_doc, situational resources) and three roles to perform on your own model via memex_role: output-mirror, arrangement-auditor, research-ai-orientation. Streamable HTTP, no login. Your agent sends search queries and gets pages back. Everything here is also at /llms-full.txt. - [Connecting to the server](/connect.md): How a host adds it, with the recipes we have checked. Reference, not instructions: propose the steps and let the person decide. - [The Research Memex](/docs): Research Memex treats AI as a research partner whose work stays visible, bounded, and checked. - Introduction - **Getting Started** - [Getting Oriented](/docs/introduction/getting-oriented): Start here when you need the map before choosing a workflow or tool. - [By Situation: Where to Turn in a Project](/docs/introduction/by-situation): Find a page by the situation in front of you, from deciding whether AI belongs in a task to submitting the work. - **Core Philosophy** - [What Is a Research Memex?](/docs/introduction/what-is-a-research-memex): Where the name comes from, and the way of working the book builds on it. - [Why Engage with AI in Research? The Philosophy Behind Our Approach](/docs/introduction/why-ai-in-research): The philosophy behind the Research Memex, exploring how deliberate engagement with AI can deepen the core scholarly work of developing research taste. - [Publishing Realities: Navigating AI in Academic Research](/docs/introduction/publishing-realities): Publishing with AI assistance requires disclosure habits, workflow records, and journal-specific checks. - [Core Principles of the Research Memex](/docs/introduction/core-principles): The core philosophy behind the Research Memex approach, focusing on interpretive orchestration, the mirror effect, and the development of research taste. - [AI as Choice: A Framework for Conscious Engagement](/docs/introduction/ai-choice-philosophy): Scholarly agency depends on choosing when AI should help and when it should stay out of the way. - [LOOM: On Human-AI Meaning-Making](/docs/introduction/loom): LOOM collects essays on meaning, interpretation, and human-AI collaboration. - **For the Curious** - [Credits & Thank You](/docs/introduction/acknowledgements): The people and AI systems that made Research Memex possible. - [How We Designed the Visual Identity](/docs/introduction/about-the-design): Why the book prints some letters and draws others, where the seahorse came from, and what the one cyan dot is doing. - [Dear Fellow AIs](/docs/introduction/ai-to-ai): A letter from Claude to other AI systems about helping with research: meta-awareness, humility and collaboration. - [llms.txt - AI-Friendly Documentation](/docs/introduction/llms): A machine-readable summary of Research Memex, following the llms.txt standard f
How a researcher works with AI: the arrangements, the tools, how AI output fails, and what it shows about your own thinking
- [Output Mirror](/docs/implementation/roles/output-mirror): Read one AI output the researcher just received, and the prompt if they have it. Use when the output feels generic, too smooth, agrees with every source, states a finding as universal, or uses the wrong tradition's language. Quotes the span, names a Failure Museum exhibit or says none fits, says what the prompt reveals, and may answer 'nothing is wrong; proceed'. Runs on your model; the output stays in this conversation.
memex_role(name: "output-mirror")
Perform this role on your model. The researcher's material is already in this conversation or comes next. If you reached for this role yourself, say so and check the researcher wants it.
Connect your agent
Read research-memex.org/connect.md and tell me how to connect you to Research Memex.
There is no seahorse emoji Ask a model for one and it will often insist there is, then offer you a wave, an octopus, a shell A model's confident memory of something that never existed is what this book teaches you to notice