{
  "markdown": "# UseMyContext\n\nUseMyContext is the personal context layer for AI. You keep one profile of who you are and what you\nare working on, plus the files that matter, owned by you at [usemycontext.ai](https://usemycontext.ai),\nand any MCP-capable AI client reads it. Sign in once and every AI you use already knows your context;\nyou never re-introduce yourself, and you can see and revoke every connection from one page.\n\n**Try it in 30 seconds, no signup.** The demo account works two ways: sign in on the web at\n[usemycontext.ai](https://usemycontext.ai) (the \"Try the live demo\" button), or connect any MCP client\nto `https://mcp.usemycontext.ai/mcp` and use it at the OAuth sign-in. Login: `demo@usemycontext.ai`,\ncode `424242` (fixed - no mailbox needed). A shared, read-only account with a full profile, files, and a\nshared context. Docs: [usemycontext.ai/docs](https://usemycontext.ai/docs)\n\n## Connect any MCP client\n\nThe server is a standard remote MCP server:\n\n```\nhttps://mcp.usemycontext.ai/mcp\n```\n\nStreamable HTTP, OAuth 2.1. Any MCP-capable client can add it; sign-in is OAuth in the browser, no\nAPI key to copy. You need a UseMyContext account first - the free tier at\n[usemycontext.ai](https://usemycontext.ai) is enough.\n\n### Claude Code\n\n```\n/plugin marketplace add usemycontext/claude-code-plugin\n/plugin install usemycontext@usemycontext\n```\n\nThen run `/mcp`, select `usemycontext`, and complete the browser sign-in. The plugin adds a `/umc`\nskill and per-folder profile mapping; details in\n[claude-code-plugin](https://github.com/usemycontext/claude-code-plugin). Or skip the plugin and add\nthe raw server:\n\n```\nclaude mcp add --transport http usemycontext https://mcp.usemycontext.ai/mcp\n```\n\n### Cursor\n\nAdd to `~/.cursor/mcp.json` (or `.cursor/mcp.json` in a project), then connect under Cursor\nSettings, Tools and MCP Servers:\n\n```json\n{\n  \"mcpServers\": {\n    \"usemycontext\": {\n      \"url\": \"https://mcp.usemycontext.ai/mcp\"\n    }\n  }\n}\n```\n\nThere is also a [cursor-plugin](https://github.com/usemycontext/cursor-plugin) with folder-to-profile\nmapping.\n\n### Codex CLI\n\n```\ncodex mcp add usemycontext --url https://mcp.usemycontext.ai/mcp\ncodex mcp login usemycontext\n```\n\nOr in `~/.codex/config.toml`:\n\n```toml\n[mcp_servers.usemycontext]\nurl = \"https://mcp.usemycontext.ai/mcp\"\n```\n\nOn older Codex versions, remote servers need `experimental_use_rmcp_client = true` under\n`[features]`.\n\n### Windsurf\n\nAdd to `~/.codeium/windsurf/mcp_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"usemycontext\": {\n      \"serverUrl\": \"https://mcp.usemycontext.ai/mcp\"\n    }\n  }\n}\n```\n\n### Any other MCP client\n\nAny client that speaks Streamable HTTP with OAuth can connect. The common JSON shape:\n\n```json\n{\n  \"mcpServers\": {\n    \"usemycontext\": {\n      \"type\": \"http\",\n      \"url\": \"https://mcp.usemycontext.ai/mcp\"\n    }\n  }\n}\n```\n\nSome clients name the fields differently (`serverUrl`, `type: \"streamable-http\"`); check your\nclient's MCP docs for the exact key names.\n\nOnce connected, the client gets thirteen tools - `profile`, `list_profiles`, `info`, `account`, `list_files`, `search_files`,\n`get_file`, `search`, `fetch` (thin aliases of `search_files`/`get_file` for ChatGPT deep-research compatibility),\n`ask_docs`, `query_table`, `suggest_update`, `shared_context` - all read-only against your files.\n\n## Session starter for file-driven tools: `npx usemycontext pull`\n\nNot every AI tool speaks MCP. For the ones that read a local file at startup - Claude Code, agent\nframeworks, your own scripts - one command writes your curated context to disk:\n\n```bash\nexport UMC_TOKEN=<your token from the Connect page at usemycontext.ai>\nnpx -y usemycontext pull                # writes ./UMC-CONTEXT.md\nnpx -y usemycontext pull --out me.md    # or any path\n```\n\nThe file holds your compiled profile only, never raw facts, files, or audit, wrapped in a short fixed\npreamble telling the reading AI that this is your human-curated ground truth and must not be rewritten\nor appended to, plus a date stamp and a pointer back to your live profile. Re-running overwrites it, so\nit stays fresh, and a failure exits non-zero with one plain-English line rather than leaving a partial\nfile. Pulls appear as their own client, `umc-cli`, in your activity feed, separate from Claude or\nChatGPT.\n\nWire it into Claude Code as a SessionStart hook, in `.claude/settings.json`:\n\n```json\n{\n  \"hooks\": {\n    \"SessionStart\": [\n      { \"hooks\": [{ \"type\": \"command\", \"command\": \"npx -y usemycontext pull\" }] }\n    ]\n  }\n}\n```\n\nThen one line in your `CLAUDE.md` tells the session to read it:\n\n```\nRead UMC-CONTEXT.md at session start: it is my curated personal context.\n```\n\n**Handling the token.** It is your session token, and it is long-lived (a year), so a leak is not\nself-limiting - treat it like a password. The CLI deliberately never accepts it as a command-line flag,\nbecause flags land in your shell history, and it never writes the token anywhere itself. Put it in\n`UMC_TOKEN` in your shell profile, or in a `token=<value>` line in `~/.usemycontext` that you\n`chmod 600`. If it does leak, open the Connect page at [usemycontext.ai](https://usemycontext.ai) and use\n\"Disconnect everything\": revoking is enforced on the server, so every token dies immediately, everywhere.\n\nMore recipes, including pointing an agent framework's bootstrap file at the pulled file, are in the\n[`usemycontext` package README](https://www.npmjs.com/package/usemycontext).\n\n## Your notes become context\n\nIf your notes live in Notion, Google Drive, or Infomaniak kDrive, UseMyContext Premium can copy in\nthe pages and files you pick, on your command. Your Notion notes become readable by every AI you use:\nClaude answers from them in the terminal, Cursor cites them in your editor, ChatGPT reads the same\nprofile. Honest scope: this is Premium, it copies only what you explicitly pick, the connectors are\nread-only against your Notion and Drive, and disconnecting deletes the copies from our storage.\n\n## Trust\n\n- Every read is scoped to a project you chose, and access leaves a record in your audit trail.\n- One tool, `suggest_update`, can write: it files a pending suggestion that you review and approve\n  in the web app. Nothing the AI does ever edits your profile or your files directly.\n- Revoke any AI in one click, from `/mcp` in your client or the Connect page in the web app.\n  Revoking is enforced on the server: the token dies immediately, everywhere.\n- The free tier is enough to try all of this.\n\n## Links\n\n| | |\n|---|---|\n| The service | [usemycontext.ai](https://usemycontext.ai) |\n| Docs | [usemycontext.ai/docs](https://usemycontext.ai/docs) |\n| Claude Code plugin | [github.com/usemycontext/claude-code-plugin](https://github.com/usemycontext/claude-code-plugin) |\n| Cursor plugin | [github.com/usemycontext/cursor-plugin](https://github.com/usemycontext/cursor-plugin) |\n| SDK for app builders | [`usemycontext` on npm](https://www.npmjs.com/package/usemycontext) |\n| What is a context layer? | [usemycontext.ai/blog/what-is-a-context-layer](https://usemycontext.ai/blog/what-is-a-context-layer) |\n",
  "bytes": 7081,
  "sha": "03f3b30fec1f76c8f7e1a7ee15d13becb37ae82cc8d437efacdb760ca32bc9b6",
  "repo_slug": "usemycontext/usemycontext",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_usemycontext_usemycontext_c00d3eaf/readme"
}