[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-83936":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":38,"readmeContent":39,"aiSummary":10,"trendingCount":15,"starSnapshotCount":15,"syncStatus":40,"lastSyncTime":41,"discoverSource":42},83936,"markdown-memory","mworldorg\u002Fmarkdown-memory","mworldorg","A cross-platform, file-based persistent memory and prompt bridge for Claude Code, claude.ai, and Antigravity IDE. Synchronizes project passports, handoffs, and session logs in an Obsidian Vault, connecting web planning with local CLI execution. Install via: npx markdown-memory","",null,"Python",56,1,52,0,3,4,10,49.3,"MIT License",false,"main",[24,25,26,27,28,29,30,31,32,33,5,34,35,36,37],"antigravity","antigravity-ide","antigravity-skills","antrophic","claude","claude-code","claude-code-memory","claude-code-skill","claude-skills","markdown","opus-4-8","skills","workflow","workflow-system","2026-06-12 04:01:42","\u003Cdiv align=\"center\">\n\n# mm — markdown memory & workflow system\n\n**A file-based memory and prompt bridge that connects claude.ai (where you think) ↔ Claude Code (where you build) ↔ Obsidian (shared long-term memory).**\n\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-blue.svg)](LICENSE)\n![Version](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fversion-0.6.0-green.svg)\n![Type: Agent Skills](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftype-Claude%20Agent%20Skills-8A2BE2.svg)\n![Platform: Windows](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-Windows%20(primary)-lightgrey.svg)\n\n\u003C!-- TODO: record a short terminal demo (asciinema \u002F GIF) and drop it here -->\n\u003C!-- \u003Cimg src=\"docs\u002Fassets\u002Fdemo.gif\" width=\"640\" alt=\"mm in action\" \u002F> -->\n\n\u003C\u002Fdiv>\n\n---\n\n## What is this?\n\nWhen you build with Claude, your work is split across two places: you sketch ideas and plan in **claude.ai**, then do the real work in **Claude Code** in your terminal. Those two sides don't share memory, and context evaporates between sessions, between chats, and whenever the context window compacts.\n\n**mm** closes that gap. It's a set of [Claude Agent Skills](https:\u002F\u002Fdocs.claude.com\u002Fen\u002Fdocs\u002Fagents-and-tools\u002Fagent-skills\u002Foverview) plus a small PowerShell installer that gives you:\n\n- **One shared memory** for every project, stored as plain markdown in an Obsidian vault (passport, handoff, session history).\n- **A prompt bridge** so an idea worked out in claude.ai becomes a clean, self-contained task you paste into Claude Code.\n- **A simple lifecycle** — start a project, resume it, save a session, hand off to a new chat — with no copy-pasting your whole history every time.\n\nEverything is markdown and files. No database, no cloud, no lock-in.\n\n---\n\n## How it works\n\nmm spans three environments that share state through files, not magic:\n\n```mermaid\nflowchart LR\n    A[\"claude.ai\u003Cbr\u002F>(ideas, planning)\u003Cbr\u002F>mm-web-bridge skill\"]\n    B[\"Claude Code \u002F PowerShell\u003Cbr\u002F>(real work)\u003Cbr\u002F>10 mm-* skills\"]\n    C[\"Obsidian vault\u003Cbr\u002F>(shared memory)\u003Cbr\u002F>passport · handoff · sessions\"]\n\n    A -- \"self-contained prompt\u003Cbr\u002F>(copy-paste)\" --> B\n    B -- \"writes notes\" --> C\n    C -- \"passport + handoff\u003Cbr\u002F>loaded into Project Knowledge\" --> A\n    C -- \"read on resume\" --> B\n```\n\n- **claude.ai side** is a single skill (`mm-web-bridge`): it challenges your idea, checks fast-moving tech against the live web, and composes the prompt you'll run in Claude Code.\n- **Claude Code side** is 10 `mm-*` skills that do the work and keep the vault up to date.\n- **Obsidian** is just the folder where the markdown lives. Obsidian indexes it for you; mm reads and writes the files directly.\n\n---\n\n## Quick start\n\n### Prerequisites\n\n- [Claude Code](https:\u002F\u002Fdocs.claude.com\u002Fen\u002Fdocs\u002Fclaude-code) CLI\n- Python 3 (installed on macOS\u002FWindows by default)\n- Node.js (v16+ for `npx` installer)\n- An Obsidian vault (any folder works; Obsidian itself is optional but recommended)\n\n### [NPM \u002F npx Installation (Recommended)](#npm-npx-installation)\n\nThe easiest way to install and configure the entire system across **Windows, macOS, and Linux** is using `npx`:\n\n```bash\nnpx markdown-memory\n```\n\nThis single command will:\n1. Clone the repository into `~\u002F.markdown-memory\u002F` (or update it if already cloned).\n2. Set the `MM_REPO_ROOT` environment variable in your shell profile (`.zshrc` \u002F `.bashrc` on Mac\u002FLinux) or Windows Registry.\n3. Automatically register all skills (creating NTFS junctions on Windows or symlinks on macOS\u002FLinux) under `~\u002F.claude\u002Fskills\u002F`.\n\n*To update the skills and codebase at any time, simply run `npx markdown-memory` (or `\u002Fmm update` inside the agent session) again!*\n\n---\n\n### Manual Installation (Alternative)\n\n```bash\n# 1. Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002Fmworldorg\u002Fmarkdown-memory.git\ncd markdown-memory\n\n# 2. Register skills:\n# On Windows (PowerShell):\npowershell scripts\u002Fregister-skills.ps1\n# On macOS \u002F Linux (or Windows with Python):\npython3 scripts\u002Fregister-skills.py\n```\n\n---\n\n### Post-Installation Setup\n\n1. **Personalization**: Restart Claude Code (skills load at session start), then run the setup command:\n   ```bash\n   \u002Fmm setup\n   ```\n2. **Create Passport**: Navigate to your project folder and initialize it:\n   ```bash\n   \u002Fmm new\n   ```\n\n### Connect the claude.ai side\n\n1. Create a Project in claude.ai for your work.\n2. Add your project's `passport.md` and `handoff.md` to **Project Knowledge**.\n3. Load the `mm-web-bridge` skill into that Project (claude.ai → Customize → Skills → Upload a skill). After `\u002Fmm setup`, upload your personalized copy from `claude-ai-skills\u002F_generated\u002Fmm-web-bridge\u002F` — the committed `claude-ai-skills\u002Fmm-web-bridge\u002F` is a template with placeholder persona.\n4. First message in a new chat:\n   > *\"Read handoff.md and passport.md, tell me where we are and suggest the next step.\"*\n\n### The loop\n\n```text\n\u002Fmm resume   →   discuss in claude.ai   →   paste prompt into Claude Code   →   \u002Fmm save   →   \u002Fmm next\n```\n\n---\n\n## Skills\n\nTen skills on the Claude Code side, plus one on the claude.ai side — and vendored external skills (see [Integrations](#integrations)).\n\n| Skill | What it does |\n|---|---|\n| `mm` | Dispatcher — short aliases for everything below |\n| `mm setup` | One-time personalization after cloning |\n| `mm-init-project` | Create a project passport + structure in the vault |\n| `mm-resume` | Load passport + dashboard + last session on start |\n| `mm-projects` | One-screen overview of all projects (read-only) |\n| `mm-bridge` | Wrap a task into a prompt framework for Claude Code |\n| `mm-handoff` | Generate the cross-chat handoff summary |\n| `mm-save-session` | Capture decisions and progress into the vault |\n| `mm-instructions` | Manage project instructions |\n| `mm-doctor` | Health checks, version sync, consistency with passport |\n| `mm-web-bridge` *(claude.ai)* | Idea partner + prompt composer in the browser |\n| `ecc-security-review` *(vendored — [ECC](https:\u002F\u002Fgithub.com\u002Faffaan-m\u002Feverything-claude-code), MIT)* | Security checklist: secrets, input validation, SQLi, auth, XSS\u002FCSRF, rate limiting |\n| `ecc-search-first` *(vendored — [ECC](https:\u002F\u002Fgithub.com\u002Faffaan-m\u002Feverything-claude-code), MIT)* | Research before coding: search existing tools\u002Flibs\u002FMCP\u002Fskills, then adopt \u002F extend \u002F build |\n\n---\n\n## Integrations\n\nmm cooperates with several external tools and bodies of work. Where it builds on someone else's project, that project keeps its own license and attribution.\n\n- **GSD** — phase-based planning framework. mm **reads** GSD state (`.planning\u002F` \u002F `.gsd\u002F`) and cooperates with it; it never writes there. *(deepest integration)*\n- **Karpathy guidelines** — four meta-principles (think before coding, simplicity first, surgical changes, goal-driven) applied across both the Claude Code and claude.ai sides. MIT. *(external Claude Code plugin — install separately, see Optional plugins below)*\n- **[context-mode](https:\u002F\u002Fgithub.com\u002Fmksglu\u002Fcontext-mode)** — in-session context optimization and continuity. See [Memory layers](#two-memory-layers) for how it relates to mm. Elastic License 2.0 (source-available). *(external Claude Code plugin — install separately, see Optional plugins below)*\n- **prompt-frameworks** — CRISPE \u002F XML \u002F PERSONA \u002F HYPOTHESIS templates used by `mm-bridge`. Inspired by [awesome-claude-prompts](https:\u002F\u002Fgithub.com\u002Flanggptai\u002Fawesome-claude-prompts), MIT.\n- **[claude-code-telegram](https:\u002F\u002Fgithub.com\u002FRichardAtCT\u002Fclaude-code-telegram)** — optional Telegram bridge, off by default. MIT.\n- **[ECC — everything-claude-code](https:\u002F\u002Fgithub.com\u002Faffaan-m\u002Feverything-claude-code)** — two skills vendored per-piece into `vendor\u002F`: `ecc-security-review` (markdown security checklist; renamed to avoid clashing with the built-in `\u002Fsecurity-review`) and `ecc-search-first` (research-before-coding workflow). The rest of ECC (AgentShield, plugin, hooks, ~246 skills) is **not** included. MIT © Affaan Mustafa. *(per-skill vendoring mechanism — see [`vendor\u002FREADME.md`](vendor\u002FREADME.md))*\n\n### Two memory layers\n\nmm and context-mode solve **different** problems and don't overlap:\n\n- **context-mode** keeps a *single session* alive — it survives auto-compaction and `--resume` by rebuilding your working state inside Claude Code.\n- **mm** keeps *long-term project history* — decisions, handoffs, and session notes that persist across sessions and across chats, in the Obsidian vault.\n\nIf you use both, context-mode already handles compaction; mm doesn't try to.\n\n### Optional plugins\n\nKarpathy guidelines and context-mode are external Claude Code plugins (not bundled in this repo). mm cooperates with them but works without them. To install:\n\n```text\n# Karpathy coding guidelines (think-before-coding, simplicity, surgical, goal-driven)\n\u002Fplugin marketplace add forrestchang\u002Fandrej-karpathy-skills\n\u002Fplugin install andrej-karpathy-skills@karpathy-skills\n\n# context-mode (in-session context optimization + continuity across compaction)\n\u002Fplugin marketplace add mksglu\u002Fcontext-mode\n\u002Fplugin install context-mode@context-mode\n```\n\n---\n\n## Cross-platform\n\nmm is cross-platform! It is fully supported on Windows, macOS, and Linux:\n- **Windows**: Use `powershell scripts\u002Fregister-skills.ps1` for PowerShell or `python3 scripts\u002Fregister-skills.py`.\n- **macOS \u002F Linux**: Use `python3 scripts\u002Fregister-skills.py` (which sets up symlinks in `~\u002F.claude\u002Fskills` and exports `MM_REPO_ROOT` in your shell profile like `.zshrc` or `.bashrc`).\n\nFor the optional Telegram bridge:\n- **Windows**: Use `powershell scripts\u002Finstall-tg-bridge.ps1`.\n- **macOS \u002F Linux**: Use `python3 scripts\u002Finstall-tg-bridge.py`.\n\n---\n\n## Conventions\n\n- **Personal data never lands in committed files.** Your name and paths live in a gitignored `mm-config.local.json` written by `\u002Fmm setup`. The committed config stays a generic template.\n- **Single source of truth for versions** — `config.version`. Per-skill `version:` fields are intentionally granular.\n- **Skills preview before they write**, and never write into GSD's `.planning\u002F` \u002F `.gsd\u002F`.\n\n---\n\n## Documentation\n\n- [`passport.md`](templates\u002Fpassport.md) — the per-project source of truth (architecture, conventions, constraints)\n- [`docs\u002FMCP-INTEGRATION.md`](docs\u002FMCP-INTEGRATION.md) — step-by-step guide to connect your Obsidian Vault via Model Context Protocol (MCP) (Recommended)\n- [`docs\u002F`](docs\u002F) — deeper guides (Telegram bridge, config loading, etc.)\n\n---\n\n## Contributing\n\nIssues and PRs welcome. New skills go in `skills\u002Fyour-skill\u002FSKILL.md` with YAML frontmatter; claude.ai skills go in `claude-ai-skills\u002F` (they are **not** junctioned into Claude Code).\n\n## License\n\nMIT — see [LICENSE](LICENSE). Integrated projects retain their own licenses (see [Integrations](#integrations)).\n",2,"2026-06-11 04:11:50","CREATED_QUERY"]