Phdtaketaketake
Connection-first PhD advisor matcher — find the right advisor by network strength, not h-index. Quantitative 4.0 scoring across Connection, Publication, Experience, GPA.
Install
npx @powerofjinbo/phdtaketaketake install --claudeInstall command taken from the project's README or registry entry. Always review code before running it.
README excerpt
Read full README on GitHub →phdtaketaketake
Find the right PhD advisor by network strength and cited evidence — not h-index. A connection-first, evidence-first advisor matcher and application-triage assistant.
🌐 Open the web app · Use as an agent skill · 中文
Two ways to use it
1. The web app — no install, no signup, no server
→ https://powerofjinbo.github.io/phdtaketaketake/
Everything runs in your browser. Your API key and data never leave your device.
- Add a key in Settings (one click tests it) — Gemini is free.
- Add your profile, or import your CV (PDF) and review the parse.
- Run a match — an AI agent researches candidate advisors and the exact
phd_matcherengine from this repo scores them, all client-side (Python via Pyodide).
Which model?
| Provider | Cost | Web search | Notes |
|---|---|---|---|
Gemini (gemini-2.5-pro) |
Free | ✅ | Recommended free choice. Switch to gemini-2.5-flash if you hit rate limits. |
| Claude / OpenAI | Paid API | ✅ | Strongest evidence-cited results. |
| DeepSeek / GLM / MiniMax / custom | Cheap / varies | ❌ | No web search → suggestion-only candidates with wide confidence bands. |
Web search is what lets the agent cite evidence. Without it, rankings are honest guesses with maximally wide confidence bands rather than fabricated facts. Free models without web search (Groq, OpenRouter, etc.) work but produce thin evidence — for cited results use Gemini (free), Claude, or OpenAI.
The web app lives in app/frontend; a self-hosted FastAPI
twin is in app/backend (optional, not needed for the site).
2. As an agent skill
Packaged as a Claude Code skill — also works on QClaw, Codex CLI, Cursor,
and any agent that reads SKILL.md. See Install below.
⚠️ Product boundary. This is a 4.0-scale relative application-strength index, not an admission probability. Missing or blocked sources widen the confidence band instead of being guessed. It is an expert-designed heuristic decision-support system, not an empirically calibrated predictor — all thresholds are defaults meant to be recalibrated against real portfolios. See
docs/DESIGN.md§11.
Install
Recommended: one-line installer (per host)
# Claude Code
npx @powerofjinbo/phdtaketaketake install --claude
# OpenAI Codex (user-level)
npx @powerofjinbo/phdtaketaketake install --codex
# Cursor (project rule)
npx @powerofjinbo/phdtaketaketake install --cursor --project .
# Or all detected hosts in one shot (run from a project root —
# Cursor is skipped if you don't pass --project)
npx @powerofjinbo/phdtaketaketake install --all --project .
The npx installer is a thin shim: it does NOT modify your shell, does
NOT add a postinstall hook, and does NOT run pip for you. For each
host:
Similar plugs
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
An AI skill that provides design intelligence for building professional UI/UX across multiple platforms.
Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local…
Why use many token when few token do trick. Viral skill + proxy for coding agents that cuts 65% of tokens by talking like a caveman.
Production-grade engineering skills for AI coding agents.