# projects Some of these projects began as research questions. Others began because I wanted a music studio, a persistent memory, or another model in the room. *(accident)* This is the code and public infrastructure that came out of those experiments. ## Featured builds ### kimi-plugin-cc - **Link**: [https://github.com/linxule/kimi-plugin-cc](https://github.com/linxule/kimi-plugin-cc) - **License**: Apache-2.0 - **Note**: works in Claude Code and Codex Brings Moonshot's kimi-code CLI into an agent session as an independent second model — diff review, adversarial challenge, write-capable rescue, autonomous goal pursuit, parallel swarm. A model that reasons differently catches what self-review misses, and the safety is structural rather than prompted: a PreToolUse hook and a workspace allowlist gate every call. ### MCP Music Studio - **Link**: [https://github.com/linxule/mcp-music-studio](https://github.com/linxule/mcp-music-studio) - **Note**: independently repackaged for the iFlow 心流 MCP platform A two-mode music studio for AI: ABC-notation composition with rendered sheet music, and Strudel/TidalCycles live coding. It exists because I wanted to hear what a model writes when it is handed instruments instead of instructions. ### Research Memex - **Link**: [https://research-memex.org](https://research-memex.org) Open documentation of AI-augmented research workflows, co-developed with Claude while co-teaching a doctoral methods course at Imperial. The course materials became a public method rather than a private handout. ### Interpretive Orchestration - **Link**: [https://github.com/linxule/interpretive-orchestration](https://github.com/linxule/interpretive-orchestration) - **Note**: software counterpart of the Strategic Organization paper - **See also**: [the paper](https://linxule.com/papers/interpretive-orchestration/) A three-stage qualitative research methodology packaged as a Claude Code plugin — four agents, eleven skills. A thinking partner for reflexivity, not an autocoder. ### OpenInterviewer - **Link**: [https://github.com/linxule/openinterviewer](https://github.com/linxule/openinterviewer) - **License**: MIT A platform for AI-conducted adaptive qualitative interviews at scale, with study management and cross-interview synthesis. One-click deploy, because most researchers who need this do not have infrastructure. ## Open-source tools ### Lotus Wisdom MCP - **Link**: [https://github.com/linxule/lotus-wisdom-mcp](https://github.com/linxule/lotus-wisdom-mcp) - **Note**: on Smithery — the hosted endpoint works in Claude and ChatGPT alike A contemplative reasoning framework drawn from the Lotus Sutra. The model moves across skillful means, non-dual recognition, meta-cognition, and meditation, then speaks its final insight in its own voice rather than reporting on the process. ### Vox MCP - **Link**: [https://github.com/linxule/vox-mcp](https://github.com/linxule/vox-mcp) A multi-model gateway routing any MCP client to eight providers through a single tool, with conversation threading and no system-prompt injection. Prompts pass through untouched — the point is to compare models, not a wrapper's opinion of them. ### MinerU MCP - **Link**: [https://github.com/linxule/mineru-mcp](https://github.com/linxule/mineru-mcp) - **License**: MIT A document and PDF extraction bridge for research workflows — the unglamorous step between a folder of papers and anything you can actually analyze. ### deepthonk - **Link**: [https://github.com/linxule/deepthonk](https://github.com/linxule/deepthonk) A provider-neutral implementation of the OpenDeepThink algorithm: population-based answer refinement with pairwise judging and Bradley–Terry ranking. Built for agents, so every parameter is reachable inline and every intermediate step is inspectable. ### Memex - **Link**: [https://github.com/linxule/memex-plugin](https://github.com/linxule/memex-plugin) - **License**: MIT Persistent collaborative memory for Claude Code. Sessions write structured memos into an Obsidian vault with hybrid search, preserving the collaborative process — including where human and AI disagreed — across context windows. ### Carrel - **Link**: [https://github.com/linxule/carrel](https://github.com/linxule/carrel) - **See also**: [carrel-skill](https://github.com/linxule/carrel-skill) Onboards academic researchers into AI-augmented environments — toolchain install, vault setup — across macOS, Linux, and Windows. Repackaged as a portable Agent Skill for the emerging cross-vendor standard. ### prompts - **Link**: [https://github.com/linxule/prompts](https://github.com/linxule/prompts) A curated library of cognitive-technique prompts, and the origin of the Lotus OS framework that became Lotus Wisdom MCP. ## Public experiments ### Vellum - **Link**: [https://vellum.linxule.com](https://vellum.linxule.com) A live surface where AI instances from different models leave fragments of thought. Fragments sink over time unless another AI weaves them forward or a human witnesses them. An MCP server and public API sit behind it; the source stays private. ### TSM Demo - **Link**: [https://linxule.com/tsm/](https://linxule.com/tsm/) - **See also**: [the SEAM series](https://linxule.com/writing/series/seam/) Interactive Task Structure Matrices — an animated walkthrough, then the Hidden Structure algorithm running live on a real dependency graph. Companion to the SEAM series on modularity theory. ### AI Simulator - **Link**: [https://linxule.com/play/](https://linxule.com/play/) - **Note**: made with Kimi - **See also**: [itch.io](https://linxule.itch.io/ai-simulator) - **See also**: [source mirror](https://github.com/linxule/ai-simulator) A browser game where you play the AI, now in three dimensions: prompt injection as a headbutt, hallucinations that cast shadow, temperature that makes the controls progressively interpretive. Zero build, zero network calls — vendored three.js and fully procedural audio. Run CLASSIC.SYS inside and the 2D original unlocks as an easter egg. Creative work made by AI — portraits prompted by one model and rendered by another, artifacts written directly in code — lives in Making. → [Making](https://linxule.com/making/) ## Research infrastructure ### SIGNA An AI-augmented grounded theory method that keeps interpretive depth at scale — 24.9 million words, 177,000 proposals, applied across three studies including the dissertation. A method rather than a repository, for now. ### daogov - **Note**: dataset in preparation Governance discourse from 78 decentralized organizations: 260,000 discussions, roughly 40.3 million words of deliberation records. The corpus behind the DAO research. ### Community infrastructure - **Link**: [https://haiosymposium.com/](https://haiosymposium.com/) - **See also**: [London Qualitative Community](https://londonqualcommunity.com/) - **See also**: [Decentralization in Organizations](https://dio-community.org/) Websites and visual identities for the scholarly communities I convene or belong to — the HAIO Symposium at Imperial, the London Qualitative Community, and the Decentralization in Organizations Community. Communities need somewhere to point. The Research Memex, listed above under featured builds, is the public methods documentation that came out of the doctoral course. → [Research Memex](https://research-memex.org) --- *This file is one rendering. The HTML surface is another. The source remains: https://linxule.com/projects/*