Live MCP Server

The asish.xyz portfolio, exposed as a Model Context Protocol server. Connect Claude Desktop, Claude.ai, Cursor, or any MCP client — and have it answer questions about my work directly.

Your MCP server URL

https://asish.xyz/api/mcp

Paste this URL into Claude Desktop, Claude.ai (web), Cursor, or any MCP client. The server is public and stateless — no authentication needed.

Connect your client

1
Open Claude Desktop → Settings → Developer → Edit Config. (macOS path: ~/Library/Application Support/Claude/claude_desktop_config.json)

Add the asish-portfolio entry to mcpServers:

{
  "mcpServers": {
    "asish-portfolio": {
      "type": "http",
      "url": "https://asish.xyz/api/mcp"
    }
  }
}

3
Restart Claude Desktop. The server appears in the connectors picker. Try: “Using the asish-portfolio MCP, what did Asish ship at CRED?”

Tools the MCP server exposes (9)

get_summaryno args

Returns a one-paragraph professional summary of Asish Madala — current role, years of experience, defining strengths, headline metrics, and links to LinkedIn / GitHub. Use this as the first call when you have no prior context about Asish. Faster and cheaper than `ask` for general 'who is this person' questions.

get_top_stories1 arg

Returns 1-3 canonical interview stories from Asish's career, each with title, role, period, summary, key actions, results, and metrics. Filter by `topic` to find stories relevant to a specific theme. Use this when a recruiter asks 'tell me about a time when...' or wants to drill into a particular kind of work (agentic AI, RAG, MCP, evaluations, shipping velocity, leadership).

topic

analyze_fit2 args

Analyzes a job description and returns a fit verdict for Asish Madala. Returns: fitScore 0-100, scoreReasoning, three matching career stories with metrics and JD alignment, two honest gaps with interview mitigation strategies, and a tailored 4-sentence cover paragraph. Takes 6-10 seconds. Use this when a recruiter pastes a JD or asks 'would he be a fit for X?'. This is the same engine that powers https://asish.xyz/fit — calling it via MCP returns the same shape, one-shot (not streamed).

jdcompanyUrl

ask1 arg

General-purpose Q&A about Asish Madala, grounded in his portfolio content. Returns a 2-4 paragraph answer plus a list of source citations (heading + snippet). Use this as the catch-all when other tools don't fit (e.g. 'how does he think about prioritization?', 'what's his leadership style?', 'tell me about his approach to evaluations'). Answers are written in Asish's voice, in the first person.

question

get_projects1 arg

Returns Asish's shippable projects (not career stories — those are in `get_top_stories`). Each project has title, tags, year, description, stack, links, and outcome metrics. Optional `tag` filter for substring search. Use this when the user asks 'what has he built' or wants concrete deliverables for a tech stack.

tag

get_experience2 args

Returns Asish's work history as a structured array of roles, most-recent first. Each role has company, title, dates, location, a 1-2 sentence summary, and 3-5 highlight bullets with metrics. Optional filters: `company` (substring match) and `since` (ISO date). Use this when the user asks about a specific company, timeline, or 'show me his recent work'.

companysince

get_educationno args

Returns Asish's formal education — Carnegie Mellon University (M.S. Computational Engineering) and Indian Institute of Technology Madras (B.Tech). Each entry has school, degree, dates, focus areas. Use when a recruiter asks about academic background or alma mater.

get_awardsno args

Returns Asish's awards, recognitions, and notable affiliations. Sparse intentionally — only items a recruiter would find credible and recent. Use when asked about academic / industry recognitions.

get_contactno args

Returns Asish's public contact information — email, LinkedIn, GitHub, location, portfolio URL. Use when the user wants to know how to reach Asish or where to find more.

Why this matters

Don't read the portfolio. Add it as a tool to your AI client and have Claude or Cursor do the work. Ask “tell me about Asish's agentic AI experience” and your client will call get_top_stories directly. Paste a JD into chat and the client calls analyze_fit — same engine as /fit, but inside Claude.

Every tool call is logged (anonymized) so I can see which clients are connecting and which tools matter most. Stateless, public, no auth — built for recruiters at AI labs to actually try.

Demo-grade by design — sketches of how I work. Production fidelity scales with the infra and compute behind it.