{
  "markdown": "# MyAITwin MCP\n\nA personal RAG database and semantic search engine you build and control\nfrom inside your AI chat. Store your knowledge, voice, and skills as you\nwork. Retrieve them in seconds, source always cited. Your AI then creates\noutput that is recognisably you, in every conversation.\n\nLive at https://myaitwin.lutolearn.com. Free during early access.\n\n## What it is\n\nMyAITwin is two things at once.\n\n**The toolbox.** A production-grade RAG database with semantic search that\nyou shape from chat. You define the structure, the types, the tags. It is\nyours, it is visible, and you are the architect of it.\n\n**The twin.** The layer on top that greets you, guides you, assesses what\nyou store, and creates output that sounds like you. It knows the difference\nbetween what you know (your knowledge) and how you say things (your skills),\nand it uses both.\n\n## Install\n\nThree steps. Under two minutes.\n\n1. Sign up at https://myaitwin.lutolearn.com/ with your email.\n2. Click the magic link, then copy your personal MCP URL from `/create`.\n3. In Claude Desktop: Settings → Connectors → Add custom connector → paste your URL.\n\nOr use the canonical OAuth-authenticated endpoint:\n\n- **URL:** `https://myaitwin.lutolearn.com/mcp`\n- **Transport:** Streamable HTTP\n- **Auth:** OAuth 2.1 with PKCE (S256) and Dynamic Client Registration\n\nRequires a client with MCP capability. Currently Claude Pro, Claude Team,\nand ChatGPT Pro.\n\n## The 19 tools\n\n**Storing knowledge**\n\n| Tool | What it does |\n|---|---|\n| `add_knowledge` | Store a typed, tagged knowledge item |\n| `add_voice_note` | Store a voice note transcript with automatic extraction |\n| `add_document` | Store a long document with automatic chunking |\n| `add_from_url` | Fetch and store a web page |\n| `add_reference_record` | Store a creation event linking knowledge and skills used |\n\n**Retrieving knowledge**\n\n| Tool | What it does |\n|---|---|\n| `search_twin` | Semantic search across all knowledge |\n| `search_for_creation` | Dual search returning skills and knowledge separately |\n| `get_by_type` | Retrieve all items of a specific type |\n| `get_by_tag` | Retrieve all items with a specific tag |\n| `list_recent` | List recently added items |\n\n**Understanding your twin**\n\n| Tool | What it does |\n|---|---|\n| `get_schema` | Overview of your types and how many items you have |\n| `get_sources` | List all source documents |\n| `find_patterns` | Surface recurring patterns across your knowledge |\n| `synthesise` | Synthesise across multiple knowledge items on a topic |\n\n**Managing your twin**\n\n| Tool | What it does |\n|---|---|\n| `get_welcome` | Session initialisation and system prompt |\n| `update_knowledge` | Update an existing item |\n| `add_schema_type` | Define a new knowledge type |\n| `update_schema_type` | Update an existing type definition |\n| `delete_knowledge` | Delete an item (destructive) |\n\nAll tools are annotated with `title`, `readOnlyHint`, and `destructiveHint`\nper the MCP spec. Of the 19: 10 read-only, 8 write (non-destructive), 1\ndestructive (`delete_knowledge`).\n\n## How it works\n\nRAG is Retrieval-Augmented Generation. It is the architecture that lets AI\nanswer using your specific knowledge rather than its training data alone.\n\nTwo layers:\n\n- **Supabase (PostgreSQL)** for structured records with types, tags, and provenance.\n- **Pinecone** for vector embeddings, so you can search by meaning rather than exact words.\n\nWhen you search, both layers work together and return results ranked by\nrelevance. Every result is cited with source and date, and tagged with\nprovenance: personal (your own thinking), organisational (from your\norganisation), or external (from someone else).\n\nThe architectural insight worth getting right:\n\n**Knowledge** is what you know. Facts, decisions, transcripts, observations.\n\n**Skills** are how you express things. Your LinkedIn voice. Your email\nstyle. Your proposal structure.\n\nExceptional output needs both. Take a meeting transcript and ask for a\nfollow-up email. The twin needs the transcript and your email skill to\nproduce something that is accurate and unmistakably yours. Neither alone is\nenough.\n\n## Security and privacy\n\n- Bearer token authentication on every request, hashed at rest.\n- OAuth 2.1 with PKCE for connector-style integration. No shared secrets,\n  no static credentials.\n- Multi-tenant data isolation: each user lives in their own namespace.\n  Other users can never read your data. Verified by a 35-check cross-tenant\n  test suite.\n- Rate limiting per tenant.\n- Append-only audit log on every tool call.\n- Prompt injection guardrails on stored content.\n- Your data is used only to provide the service. Never used to train AI\n  models. Never shared with third parties.\n- You can delete your account and all data instantly from `/create`.\n  Deletion is immediate and irreversible.\n\nPrivacy policy: https://myaitwin.lutolearn.com/privacy\nSecurity contact: security@lutolearning.com\nPrivacy contact: privacy@lutolearning.com\n\n## Distribution\n\n- **Official MCP Registry:** [`com.lutolearn/myaitwin`](https://registry.modelcontextprotocol.io/v0/servers?search=myaitwin)\n- **Anthropic Connectors Directory:** submitted, in review\n- **Listed at:** Glama, mcp.so, mcp.directory, mcpserverfinder, Hugging Face,\n  awesome-mcp-servers\n\n## License\n\nMIT. See [LICENSE](./LICENSE).\n\n## Links\n\n- Website: https://myaitwin.lutolearn.com\n- Documentation: https://myaitwin.lutolearn.com/docs\n- Privacy: https://myaitwin.lutolearn.com/privacy\n- Support: support@lutolearning.com\n\n---\n\n*MyAITwin MCP by Luto.*\n",
  "bytes": 5524,
  "sha": "9b83ac66695555a83777f11723050d4dbc3b893059095afda032e835b2e13090",
  "repo_slug": "lutolabs/mytwin-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_com_lutolearn_myaitwin_86e2c03d/readme"
}