Explore my work with your AI assistant.
This portfolio is written to be read by people and by AI agents. Connect an MCP-compatible assistant, ask about my experience, case studies and projects, and it answers from the content published here.
01 · How it works
Your assistant asks. The server answers from the portfolio.
01
Your AI assistant
Reads your question and decides which tools to call.
02
MCP server
/api/mcp
Exposes read-only tools over the published content. It doesn’t run a language model.
03
Portfolio content
Experience, case studies and projects, as published on this site, in English.
The reasoning happens in your assistant. The server only finds and returns content that is already published on this site.
02 · What you can explore
Four read-only tools.
Profile
Name, role, positioning and contact details, as published.
Professional experience
Roles, teams and responsibilities, by employer and period.
Case study list
Browse the published case studies, professional and personal, kept apart.
Case study detail
Challenges, decisions, trade-offs and outcomes, section by section.
03 · Connect
Connect in three steps.
1
Copy the endpoint
Use the MCP server endpoint to configure the connection in your assistant.
2
Add it to your assistant
As a remote MCP server. Your client’s documentation shows where. In your assistant's MCP settings, add the server URL and name the connection 'Amir Ben-Assayag Portfolio'.
3
Ask a question
About my experience, case studies or projects.
Connection status: Checking status
Checking whether the server is accepting connections.
MCP endpoint
https://amir-benassayag-portfolio.vercel.app/api/mcpAvailable once the status is confirmed.
- Access
- Read-only
- Content language
- English
Client instructions
Step-by-step instructions are published only for clients where the connection has been tested.
ClaudeNot yet verified
Steps follow the vendor’s documentation, checked on 20 October 2026. Not yet tested against this server.
Requirements
- Free, Pro, Max, Team and Enterprise plans. Free plans are limited to one custom connector.
- On Team, only Owners can add a connector, under Organization settings. On Enterprise, Owners and roles that can manage access to Libraries can.
- The server must be reachable from Anthropic’s cloud over the public internet.
Steps
- Go to Customize > Connectors, select + Add, then Add custom connector.
- Enter the name 'Amir Ben-Assayag Portfolio' and the endpoint above as the URL, then select Continue.
- Under Authentication, choose No sign in, then select Add.
- On Team and Enterprise, an Owner adds it under Organization settings > Connectors > Add > Custom > Web. Members then select Connect under Customize > Connectors.
Where it goes
The Name and URL fields of the Add custom connector dialog.Authentication
No sign in. The server needs no account, token or headers.Check that it works
Open a chat, select + at the lower left, then Connectors, and turn on 'Amir Ben-Assayag Portfolio'. Ask: Which systems did Amir work on at Intuit? Check that Claude calls a portfolio tool (it may ask you to approve it).Claude: custom connectors (Official documentation (English))
ChatGPTNot yet verified
Steps follow the vendor’s documentation, checked on 20 October 2026. Not yet tested against this server.
Requirements
- ChatGPT on the web only, not mobile.
- Business and Enterprise/Edu workspaces: admins and owners, or members an Enterprise/Edu admin has allowed (the Create plugins with MCPs permission). OpenAI’s page says Pro accounts can connect custom servers with read-only access.
- The endpoint must be a public HTTPS MCP URL.
Steps
- Open Plugins, select the plus button, then Add custom MCP server.
- Enter the name 'Amir Ben-Assayag Portfolio'. A description is optional.
- Under Connection, enter the endpoint above.
- Set authentication to none, because the server needs no sign in.
- Acknowledge the risk warning, select Create as a plugin, then review the discovered tools.
Where it goes
The Name and Connection fields of the Add custom MCP server form.Authentication
None. The server needs no account, token or headers. OpenAI’s page doesn’t name this option, so choose the form’s no-authentication choice.Check that it works
Start a new conversation, type @ and select the plugin, then ask: Which systems did Amir work on at Intuit? Confirm any prompt that appears.ChatGPT: connecting custom MCP servers (Official documentation (English))
CursorNot yet verified
Steps follow the vendor’s documentation, checked on 20 October 2026. Not yet tested against this server.
Requirements
- Cursor supports stdio, SSE and Streamable HTTP servers.
~/.cursor/mcp.jsonapplies to every project..cursor/mcp.jsonapplies to one project.- Organizations can restrict remote MCP URLs.
Steps
- Open
~/.cursor/mcp.jsonto use the server everywhere, or.cursor/mcp.jsonin one project. Create the file if it doesn’t exist. - Add the JSON below. The key under
mcpServersis the server name. - Save the file, open Customize and make sure the server is enabled.
{ "mcpServers": { "Amir Ben-Assayag Portfolio": { "url": "https://amir-benassayag-portfolio.vercel.app/api/mcp" } } }Available once the status is confirmed.
Where it goes
The key undermcpServersis 'Amir Ben-Assayag Portfolio'. The endpoint is theurlvalue.Authentication
None. Leave outheaders: the server needs no token.Check that it works
In Customize, check that the server’s toggle is on and its tools are listed. If it fails, open the Output panel (Cmd+Shift+U) and choose MCP Logs. Then ask: Which systems did Amir work on at Intuit?Cursor: Model Context Protocol (Official documentation (English))
Other MCP ClientsNot yet verified
Clients that support remote MCP servers use the same endpoint.
04 · Example questions
Start with a question.
- Which systems did Amir work on at Intuit?
- How did Amir design the Kafka Router?
- What experience does Amir have with distributed systems?
- What backend technologies has Amir used?
The MCP server returns English content only. You can ask in any language your assistant supports.
05 · Boundaries
Ask anything. Change nothing.
What it reads
- Public portfolio content: experience, case studies and projects, as published here.
- Content in English. These pages are bilingual; what the server returns is English.
What it never does
- Access private files in the code repository.
- Write, edit or delete anything.
- Host or run a language model.