get_profile
Profile
Who Harshith Nayaka L is: role, employer, location, availability, contact email and public profiles. Start here for questions about the person rather than a project.
無需輸入參數。

io.github.HarshithNayakaL由 GitHub 帳號 HarshithNayakaL 發布v1.1.0更新於 Oct 30, 2026
Read-only MCP access to Harshith Nayaka L's portfolio: profile, case studies, FAQ and guides.
發布方自己的服務:PortfolioFit
以唯讀方式存取 Harshith Nayaka L 的作品集網站,讓助理查詢其個人簡介、專案、案例研究、常見問題與 agent 技能。
Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。
把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。
閱讀 https://sourceweft.com/mcp/io.github.HarshithNayakaL%2Fportfolio/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。把它新增到你客戶端的 mcpServers 設定中。
{
"mcpServers": {
"harshith-nayaka-l-portfolio": {
"type": "http",
"url": "https://harshith-nayaka-l-portfolio.vercel.app/mcp"
}
}
}Source for harshith-nayaka-l-portfolio.vercel.app, the portfolio of Harshith Nayaka L, an AI Engineer in Bengaluru who builds AI agents, RAG pipelines and LLM applications. It holds a case study for each project, guides drawn from them, and his profile, experience and published research. One scrolling page plus case-study, guide and legal routes.
The site is also built to be read by machines: every page is prerendered to static HTML, has a markdown twin, and is mirrored by a read-only JSON API.
The visual design and the rules that keep it consistent are in
AGENTS.md; the previous design is archived on the
classic-design branch.
@tailwindcss/vite); all theme tokens live in src/index.css@fontsource: Inter Variable (body and labels), Instrument
Serif (display headings), Poppins (hero headline) — served from public/fonts, preloaded in index.html@bible-strong/avatar-core for the mascot's geometry@vercel/edge for the content-negotiation middlewareNo animation library. Scroll reveals and the page transition are CSS keyframes and IntersectionObserver; pulling in a runtime for that put the whole library in the initial bundle.
npm run build is four stages, and the later ones are the ones that matter:
vite build emits the client bundle, a second vite build --ssr emits
dist-ssr/entry-server.js, scripts/prerender.mjs walks every route and writes
real HTML plus each page's markdown twin, and scripts/build-api.mjs emits the
static JSON API. A plain vite build (npm run build:client) skips all of
that and produces an empty-shell SPA.
verify:llms takes an origin as its first argument. Give it a Vercel
deployment: vite preview runs neither the middleware nor the rewrites, and
falls back to the root page for every nested route, so it fails pages that are
fine in production.
Vercel. Build command npm run build, output dist.
middleware.ts runs at the edge and does the content negotiation, the API
route table, and JSON errors for anything under /api.vercel.json carries the rewrites and headers.public/_redirects is there for Netlify; nothing reads it on Vercel.Everything real lives in src/data/:
projects.ts — Selected Work cards, plus email / GitHub / LinkedIn.caseStudies.ts — full case studies (problem, build, pipeline diagram,
judgment calls, results, tech). The architecture diagrams are data-driven:
edit a pipeline array and the node graph re-renders.skills.ts — the Agent Skills section.seo.ts — per-route titles, descriptions and JSON-LD.onee.avatar.json — the mascot's definition.Adding or removing a project means touching projects.ts, caseStudies.ts,
and the slug sets in middleware.ts. scripts/build-api.mjs fails the build if
those drift apart, so a mismatch is caught rather than shipped.
src/components/GlassFilter.tsx holds a CONFIG block using the parameter
names from liquid-glass.ybouane.com —
blurAmount, refraction, chromAberration, edgeHighlight, fresnel,
zRadius, and so on. Tune the bar by changing those numbers; the filter graph
is built from them.
That library itself can't run here: it sizes a WebGL canvas to everything it
refracts, and this page is 12,473px tall on desktop and 19,236px on a phone,
which needs a canvas past Chromium's 16,384px limit. So the effect is rebuilt
as one SVG filter referenced from backdrop-filter. WebKit and Gecko don't
resolve filter references there and fall back to the plain blur declared
alongside it — same frost, no refraction at the rim.
og:url currently point at the
vercel.app subdomain.來源:README.md,提交 6276036
Profile
Who Harshith Nayaka L is: role, employer, location, availability, contact email and public profiles. Start here for questions about the person rather than a project.
無需輸入參數。
Projects
Every project in the portfolio with a one-line outcome, tags, status, links and its slug, optionally narrowed to one topic. Use the slug with get_case_study for the full write-up.
topicCase study
The full case study for one project: the problem, what was built, the pipeline, the engineering decisions, results, stack and sources.
slug必填FAQ
The questions and answers from the site's FAQ (hiring and availability, what the role involves, technical questions answered from the projects), optionally ranked against a question.
queryAgent skills
The published agent skills: what each one does, the method behind it, and its repository. Pass a name for one skill.
nameAnswer a question
Answers a question about Harshith Nayaka L or his work by quoting the site's own published answer, with its source URL. Returns "no published answer" rather than composing one.
question必填已滿足適用檢查項的 90%
該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。
請在儲存庫或註冊中心條目中宣告授權條款。

fit.portfolio官方發布者
PortfolioFit 是遠端 MCP 伺服器,讓助理依投資指令核對標準化投資組合,並計算配置限制與統計指標。

dev.rye官方發布者
遠端 MCP 端點,提供 Cameron Rye 的部落格、作品集專案、簡介、閱讀清單與網站即時統計,以及聯絡與訂閱工具。
由 GitHub 帳號 shlokbam 發布
瞭解AI工程師Shlok Bam:專案、經驗、技能、專利。即時、唯讀。AI 翻譯

由 GitHub 帳號 HarshithNayakaL 發布
唯讀 MCP 伺服器,提供某作品集網站 JSON API、MCP、A2A 與 NLWeb 的開發者文件。

由 GitHub 帳號 kobimantzur 發布
唯讀 MCP 伺服器,在本機登入 IBI(以色列券商)NextGen 帳戶,讓 AI 助理讀取持倉、現金與損益。
npm installnpm run dev # http://localhost:5173npm run build # type-check, client build, SSR build, prerender, APInpm run preview # serve the production buildnpm run lint # type-check onlynpm run verify:negotiation # 25 client shapes get the right representationnpm run verify:onee # every mascot animation is reachablenpm run verify:llms <origin> # needs a Vercel deployment, not npm run preview