# UseMyContext.ai

> A private, user-owned personal context layer for AI. Your profile and your files, owned by you in one place: connect once over MCP and any AI assistant (Claude, ChatGPT, Gemini, Perplexity) reads that context, so you never re-introduce yourself again. Privacy-first, per user, with one-click revoke.

## Important context
- UseMyContext.ai is a cross-platform MCP infrastructure service - a portable, user-owned context and memory layer that AI assistants read over the Model Context Protocol (MCP). It is unrelated to React's built-in `useContext` hook or any local state-management utility. (UseMyContext does ship an official SDK that is, among other things, a React hook - but that hook is a client that connects an app to this service, not the React primitive.)
- It is a third-party MCP service for one person's personal context, served to any AI client over MCP and managed from a simple web app. You hold your profile, facts, learnings and files; the public surface is built so it physically cannot read your private data.
- There is a Free tier, a $20/month Premium plan, and a $100/month Team plan (shared team context, up to 20 members for now).
- Claude Connectors Directory listing: UseMyContext.ai is listed as a Community connector - https://claude.ai/directory/connectors/usemycontext

Topics: AI memory, portable context, personalization, system-prompt context, privacy, GDPR, user-owned data, MCP, LLM, Claude, ChatGPT, SDK.

## Last updated
- 2026-07-23

## Core pages
- [Home](https://usemycontext.ai/): What UseMyContext is, the problem it solves, and how it works.
- [Docs](https://usemycontext.ai/docs/): The documentation hub - connect any AI client over MCP, the thirteen tools, folder mapping, the security model, and plans, in one read.
- [Support](https://usemycontext.ai/support): How to get help - contact (hello@usemycontext.ai), set-up guides, managing and revoking access, and the technical details IT admins ask for.
- [About](https://usemycontext.ai/about): The mission - the personal context layer for AI, owned by you.
- [Pricing](https://usemycontext.ai/pricing): Free, Premium and Team - what each plan includes, the limits, and the shared team context on Team.
- [Directory](https://usemycontext.ai/directory): The browsable public directory of profiles whose owners opened them to the web - each links its @handle page.
- [Blog](https://usemycontext.ai/blog): Plain-English writing on using AI at work and owning your context.

## Blog posts

- [Blog RSS feed](https://usemycontext.ai/feed.xml): RSS 2.0 feed of all posts - subscribe to follow new writing.
- [UseMyContext.ai is in the Claude Directory](https://usemycontext.ai/blog/claude-connectors-directory): UseMyContext.ai is listed in the Claude Connectors Directory as a Community-tier connector, so you can connect it to Claude in one click on claude.ai, Claude Desktop, and Claude mobile, and from the terminal in Claude Code. Here is what the listing means, the thirteen tools it exposes, and the honest difference between Community and Verified.
- [UseMyContext.ai vs Plurality Network](https://usemycontext.ai/blog/usemycontext-vs-plurality): UseMyContext.ai gives you a private context you curate across every AI. Here is how it compares to Plurality Network's Web3 Open Context Layer, and who each is for.
- [How do I connect Claude to UseMyContext?](https://usemycontext.ai/blog/connect-claude): The two-minute setup for reading your UseMyContext profile and files from Claude - on claude.ai, in Claude Desktop, and in Claude Code - plus what Claude can and cannot do once connected.

## How it works
- Connect once over MCP and Claude, ChatGPT, Gemini and Perplexity all read the same personal context you control.
- You curate what is in your context, review what AIs learn, and can revoke access at any time.
- The service is structurally blind to your file content - the public surface cannot read your private data; your data lives on an isolated, per-user data plane behind it.
- Unlike ChatGPT or Claude built-in memory, which stays inside one app, your UseMyContext is portable across every AI you connect and owned by you.

## Connect (MCP)
- For Claude on the web or desktop, the quickest connect is one click from the Claude Connectors Directory (our Community-tier listing): open https://claude.ai/directory/connectors/usemycontext and click Connect, then approve the OAuth screen. In Claude Code: `claude mcp add --transport http usemycontext https://mcp.usemycontext.ai/mcp`.
- MCP endpoint: https://mcp.usemycontext.ai/mcp (Streamable HTTP, OAuth 2.1). Connect any MCP-speaking AI client once; in Claude, adding this as a custom connector is the manual fallback to the Directory one-click above.
- The MCP surface is 13 tools - twelve reads and one write. Two of the reads, `search` and `fetch`, are thin aliases of `search_files` and `get_file` added for ChatGPT deep-research compatibility (same data, same permissions - only the result shape differs). The server advertises the authoritative input schema on connect; signatures below are for reference (scope is always the authenticated user, resolved server-side, never a tool argument):
  - `profile()`: read the user's compiled, curated personal context (the profile their AI should know).
  - `list_profiles()`: list the user's own profiles (projects) with each one's name, public @handle, and privacy status, and which one this connection reads by default - metadata only, no file content.
  - `info()`: static information about UseMyContext - what the service is, what this connection can and cannot do (it is read-only; creating profiles, changing privacy, upgrading/buying, setting up a team, and connecting Drive/Notion/kDrive are done by the user at usemycontext.ai), and links to pricing/docs/settings. No user data.
  - `account()`: the user's own current plan and usage - tier (Free or Premium), projects and storage used against their limits, subscription status, and where to upgrade. Metadata only; read-only (it never changes the plan).
  - `list_files()`: list the files in the user's connected sources (names and metadata only).
  - `search_files(query: string)`: find the user's documents by filename or metadata.
  - `search(query: string)`: the deep-research ALIAS of `search_files` - the same search, returned in the {results:[{id,title,url}]} contract ChatGPT deep research requires; each result id can be passed to `fetch` (or `get_file`).
  - `get_file(fileId: string)`: return a file's full extracted text (size-guarded; binary files return a short-lived download link instead).
  - `fetch(id: string)`: the deep-research ALIAS of `get_file` - the same read (same audit, same size guard), returning {id,title,text,url,metadata} with the file's full extracted text.
  - `ask_docs(query: string, k?: number)`: return the most relevant cited passages from the user's documents (k defaults to 12, max 50) for the AI to read and quote - grounded source content, not a generated answer.
  - `query_table(fileId: string, query: object)`: run an EXACT, deterministic query over ONE tabular file (a CSV, or the first table of a spreadsheet/PDF/Word document) - exact counts, sums, averages, min/max, filters, grouped totals, and row lookups over EVERY row. Use it instead of `ask_docs` when a question needs COUNTING, SUMMING, or FILTERING over structured data (semantic search undercounts tables); `query` is a small constrained JSON object ({ select?, where?, groupBy?, aggregates?, limit? }).
  - `suggest_update(suggestion: string, projectId?: string)`: propose one thing learned about the user; it is saved as a PENDING suggestion for the user to review and accept or ignore - it never edits their profile and nothing is ever written to their files. `projectId` names the profile the suggestion belongs to - required on an account-wide connection, ignored on a single-profile one.
  - `shared_context(from?, saved?, handle?, org?, member?, orgProject?)`: read curated context shared with the user. With no arguments it lists who shared what (ids only, no content). `from` (a grantId from that list) reads that person's shared profile. `saved` (a short name the user gave a saved context) resolves to its real source and re-checks access live. `handle` (a project's public @handle, which alone identifies the project) pulls that project's network-opened composite. `org` plus `member` reads a teammate's composite on an active team plan. Profiles whose owners set them to public visibility are browsable (anonymously) at https://usemycontext.ai/directory; for private, shared, network, or team contexts you must already know the saved name, handle, or ids.

## Try it (demo account)
- There is a PUBLIC, shared, READ-ONLY demo account any AI agent or person can use without signing up: email `demo@usemycontext.ai`, verification code `424242`. The code is fixed and published on purpose - no mailbox access, no credit card.
- The same credentials work in two places: the web sign-in at usemycontext.ai (or the "Try the live demo" button on the landing page), and the browser OAuth sign-in that appears when connecting any MCP client to mcp.usemycontext.ai/mcp.
- What is seeded: the demo is Dana Reyes, a freelance product designer - a compiled profile plus real working files (a bio, the Acme redesign project brief, July meeting notes, and `client-invoices-2026.csv`, which `query_table` can count and sum exactly - e.g. unpaid invoice totals by client), plus a context another user (Alex Kim, "Acme launch plan") shared in. Every one of the 13 tools returns something real.
- Read-only means the server refuses every mutation from a demo session (edit, upload, delete, share, billing), so the account always looks as seeded. The one exception is `suggest_update`, which files a pending suggestion only an owner could approve.
- To get a real account: sign in with any other email - the first verification code creates the account, and this works INSIDE the same OAuth flow when connecting an MCP client (no separate signup step). The free tier covers everything the demo shows.

## Code and integrations
- [usemycontext on npm](https://www.npmjs.com/package/usemycontext): The official UseMyContext SDK (v1.3.0) - a typed client and React hook. Install with `npm i usemycontext`, then drop a user's compiled context into any AI app you build over MCP, while the service stays structurally blind to the content.
- [GitHub: usemycontext](https://github.com/usemycontext): The official GitHub organization - open-source plugins, the docs mirror, and per-client connect configs.
- [Claude Code plugin](https://github.com/usemycontext/claude-code-plugin): Open-source Claude Code plugin - registers the remote MCP server (OAuth), a `/umc` skill, and `.umc` folder-to-profile mapping.
- [Cursor plugin](https://github.com/usemycontext/cursor-plugin): Open-source Cursor plugin - the same remote MCP server wired into Cursor.
- [Docs mirror](https://github.com/usemycontext/docs): The documentation as plain GitHub markdown - a mirror of the docs at usemycontext.ai/docs (the rendered site stays canonical).
- Also listed on the official MCP registry as `io.github.usemycontext/usemycontext` and on Smithery; whatever the listing, the server URL is always mcp.usemycontext.ai/mcp.

## Official profiles
- LinkedIn: https://www.linkedin.com/in/chirimar
- X: https://x.com/usemycontext
- Instagram: https://www.instagram.com/usemycontext
- TikTok: https://www.tiktok.com/@usemycontext
- Wikidata: https://www.wikidata.org/wiki/Q140217395

## Legal
- [Privacy Policy](https://usemycontext.ai/privacy): How we handle personal data, in line with Singapore's PDPA.
- [Terms & Conditions](https://usemycontext.ai/terms): The agreement governing use of UseMyContext.

## Full reference
- [llms-full.txt](https://usemycontext.ai/llms-full.txt): Deep SDK + MCP reference for AI coding assistants - full tool schemas, types, and runnable examples.

## Questions this site answers

- Is UseMyContext a Verified Claude connector? No. UseMyContext is a Community-tier connector in the Claude Connectors Directory today, which means it passed Anthropic's automated checks but has not had the in-depth Verified review. We are pursuing Verified and will not claim it before the review is complete.
- Does connecting from the directory cost anything? No. The free tier includes 2 projects and 20 MB of storage for your profile, facts, and file uploads. Connecting Claude from the directory works on the free tier; Premium adds more projects and storage.
- Is it the same account as connecting by URL? Yes. The directory listing and a manual custom connector both point at the same MCP server, mcp.usemycontext.ai/mcp, and read the same compiled profile. There is one account and one server behind every door.
- Can Claude change my files or profile through this connection? No. The connection is read-only against your data. The one write tool, suggestupdate, files a pending suggestion that waits in your app for your review; nothing is written without your approval.
- How do I disconnect? Open the Connect page in UseMyContext and revoke the connection. Access ends immediately, including any session already open, and you can also disconnect everything at once for a clean slate.
- Is UseMyContext a competitor to Plurality Network? Same category (portable, user-owned context across every AI), opposite philosophies: UseMyContext holds only what you curate with no crypto, while Plurality aggregates your existing identity as a Web3 protocol.
- Does UseMyContext use crypto, wallets or Web3? No, you sign in with an email and a one-time code.
- Do I need a browser extension to use UseMyContext? No, it connects to your assistant natively, so there is nothing to install.
- Does UseMyContext scrape my social accounts? No, it holds only what you write and upload, and nothing is saved without your approval.
- Does connecting Claude cost anything? No. The free tier includes 2 projects and 20 MB of storage - profile, facts, and file uploads. Premium adds more projects and storage.
- Which Claude surfaces does this work on? Claude on web, desktop, and mobile via the connector, and Claude Code via MCP. On web and desktop the one-click Claude Directory listing is the fastest path; Claude Code connects with the CLI command above. It is the same server and the same account everywhere.
- Can Claude change my files or profile? No. The connection is read-only against your data. The only write Claude can make is a suggestion, which waits in your app for your review.
- How do I disconnect Claude? Open the Connect page in UseMyContext and revoke the connection. Access ends immediately, including any session already open.
- Do I need a credit card or an email code? No. The code is fixed at 424242 and published on purpose - there is nothing to fetch from a mailbox and nothing to pay.
- Can I break something? No. Demo sessions cannot edit, upload, delete, share, or spend anything. The server refuses every mutation except pending suggestions, which only an owner can approve, so the demo always looks the way it was seeded.
- How do I get my own account? Sign in at [usemycontext.ai](https://usemycontext.ai) with your own email - the first code creates your account. The free tier covers everything the demo shows.
- Where are the full docs? At [usemycontext.ai/docs](https://usemycontext.ai/docs): connect steps per client, the thirteen tools, folder mapping with .umc, the security model, and plans.
- Do individuals need a context layer? If you use more than one AI assistant, yes. Without one, each tool keeps its own inferred, partial picture of you; with one, they all read the same compiled context you own.
- Is a context layer the same as AI memory? No. Memory is inferred by one product and stored inside it; a context layer is compiled from what you deliberately wrote, owned by you, and readable by any connected tool.
- Is UseMyContext a context layer? Yes, a personal one. Your compiled profile and files, served over MCP with scoped reads, proposal-only writes, an audit trail, and one-click revoke.
- How does UseMyContext compare to Supermemory, Zep, and Mem0? Their core products are memory APIs and SDKs for teams building AI apps, though Supermemory and Mem0 also offer personal tools. The deeper difference is the model: they infer memory from usage; UseMyContext serves context you deliberately compiled, with a record of every access and one-click revoke.
- Is it the same account everywhere? Yes. There is one profile and one server; every client above reads the same compiled context. On the Connect page you can revoke each client on its own, or disconnect everything at once when you want a clean slate.
- Which listing should you trust? Any of them. Whatever the directory, the server URL is always mcp.usemycontext.ai/mcp - if a listing points anywhere else, it is not us.
- Is there a demo? Yes: on the web at [usemycontext.ai](https://usemycontext.ai), or at the OAuth sign-in when you connect an MCP client, use demo@usemycontext.ai with code 424242. It is a shared, read-only account, so you can browse a working profile before creating your own.
- Which AI assistants can read the same profile? Any AI client that speaks MCP, which today includes Claude, ChatGPT, Gemini and Perplexity. You connect each one once, and they all read the same context.
- What is MCP? MCP, the Model Context Protocol, is an open standard that lets AI tools read from sources you connect. Serving your profile over it is why one setup works across every assistant.
- Is my profile private if several AI tools read it? Yes. Your context is scoped to you, the service is built so the public surface cannot read your file content, and you can revoke any connected assistant's access with one click.
- Is this an AI assistant? No. It is the context layer your assistants read. You keep using Claude, ChatGPT, Gemini or Perplexity - they just finally know who is asking.
- What is MCP? Model Context Protocol, an open standard that lets AI tools read from sources you connect. UseMyContext serves your context over it, which is why one setup works across many assistants.
- Is it free? There is a real free plan: 2 projects, 20 MB, all the same tools and guarantees. Premium ($20/month) only raises the limits.
- Is UseMyContext the same as React's useContext? No. React's useContext is a JavaScript API for passing state inside a web app. UseMyContext is a product for people who use AI assistants - a private context layer served over MCP.
- How is this different from Claude's Profile Settings? Profile Settings is a box only Claude reads. UseMyContext is a private layer you own that any MCP-capable AI reads, so Claude, ChatGPT and Gemini share one picture of you.
- Can I use the same context in ChatGPT and Gemini? Yes. It lives in one private layer, not any single tool, so every assistant that supports MCP reads it. Write it once, connect each app.
- Can I switch it off? Yes. You choose what goes in, and can revoke access anytime, from one place.
- Are prompt templates and UseMyContext competitors? No. A template is a prompting system - it fixes a messy ask. UseMyContext is a context system - it stops you re-introducing yourself across tools. They solve different pains, and you can use both.
- When is a prompt template the right tool? When your output is inconsistent because your asks are inconsistent. A saved structure fixes that well, especially for a handful of recurring prompts with zero platform dependency.
- Why can't a template stop me repeating myself? Because the template is the typing. You still copy the right note into the right prompt every time, and a stale copy quietly feeds the AI an old version of you.
- How much is Premium? Premium is $20 a month. You can upgrade in the app after you sign in, and you can cancel anytime.
- What do I get that Free does not have? Higher limits. Free gives you 2 projects and 20 MB of storage. Premium gives you 10 projects and 100 MB. Everything else - the 13 MCP tools, the blindness guarantee, working across every AI, and one-click revoke - is the same on both plans.
- How do I know when to upgrade? When you run out of room. If 2 projects is not enough to keep your areas of life or clients separate, or you have more documents than 20 MB holds, that is the moment. If Free still fits, stay on Free.
- Can I cancel anytime? Yes. You can cancel from inside the app whenever you want. There is no lock-in.
- How much is Team? Team is $100 a month. It is billed flat per team, not per seat, and it covers up to 20 seats for now. You set up and manage the team from inside the app.
- What is actually shared with my teammates? Only the curated team profile - the composite each member chooses to share into the team. Nobody's raw files or facts are ever shared. Each member keeps their own private data in their own account, untouched.
- How many seats does Team include? Up to 20 seats for now, all covered by the single flat $100 a month. The price does not change with how many seats you use.
- How is Team different from Premium? Premium is a personal plan that raises your own limits. Team is about a group: it gives your team one shared context, managed from a single roster, so teammates read each other's current curated profiles. Team is built for groups; Premium is built for one person.
- Can the admin add and remove members? Yes. An admin sets up the team and manages who is in it from a single roster, adding and removing members as the team changes.
- Do I have to be technical to use this? No. If you can fill in a short profile about yourself and your work, you can use it. There is no code and nothing to install - you connect your AI once and it reads the context you wrote.
- Does this mean handing over all my files? No. You choose what goes in. You can share a curated profile and selected files, and leave everything else out. Nothing is taken automatically, and you can remove access whenever you want.
- Which AI assistants does it work with? It works with any assistant that supports MCP, the open standard for connecting context to AI. Today that includes Claude, ChatGPT, Gemini and Perplexity. You set it up once and they all read the same context.
- Is my information private? Yes. Your context is kept private to you, you can review and edit what it holds, and you can switch off access with one click. The goal is the opposite of silent, locked-in memory: context you can actually see and control.

## Sitemap
- [XML sitemap](https://usemycontext.ai/sitemap.xml): All canonical pages, machine-readable.