Finds the sources, keeps what it read, and answers with a citation that still opens next quarter.
Built for: The analyst starting a market, vendor or literature review that will outlive the deck it is going into.
The button opens the checkout, where the 5 servers and 5 skills are listed one by one. Every member is free, so the checkout lists them and asks for no card.
The reason to buy a stack rather than five listings: each line below needs more than one member connected at the same time.
Search the live web through Tavily and Perplexity, and keep a dated copy of the pages worth citing with the Crawl4AI skill.
Find academic papers across OpenAlex, Semantic Scholar, CrossRef, PubMed and arXiv, and keep the citation with the claim.
Pull the transcript of a talk or interview from YouTube and quote it accurately rather than from memory.
Keep the review's working files in one Drive folder and its tables in CSV, so the second week of a review builds on the first.
What each part contributes, and why it was picked over the obvious alternative. Prices and permissions are read from the listings, so nothing here can disagree with the catalogue.
Web search built for retrieval — results with the context a citation needs attached, instead of ten blue links.
The answer-with-sources pass, useful for orientation on a topic before the systematic reading starts.
Searches Drive and reads file contents, which is where the half of the evidence that never was on the web already lives. The same server can also create, edit and share files; the read-only sign-in below keeps it to reading.
Downloads the transcript of a talk or interview, so a quote from a recording can be checked instead of remembered.
Explores the dataset a source publishes alongside its claims, with streaming for the files too large to open.
Searches up to seven academic sources at once, which is the difference between a literature review and whatever the first search engine returned.
For JavaScript-heavy sources and multi-URL crawls with reusable extraction schemas, where a single scrape call is not enough.
Ingests video and audio and makes it searchable, which is what turns twelve recorded interviews into evidence you can quote.
Grades health evidence by tiered trusted sources, for reviews that touch clinical claims and must not treat a blog and a trial as equals.
The readout at the end: the deck built from the notes, rather than a separate week of copying findings into slides.
Every member is free, so the monthly figure is $0 whatever you do with the stack: nothing here is billed by the call, and only adding a paid member could change it.
| Part | What you are paying for | Monthly, as quoted |
|---|---|---|
| Tavily Search | Free | — |
| Perplexity Ai | Free | — |
| Google Drive | Free | — |
| Youtube Transcript | Free | — |
| Csv Explorer | Free | — |
| Skills | ||
| Paper Search Pro | Free · context cost only | — |
| Crawl4AI | Free · context cost only | — |
| VideoDB | Free · context cost only | — |
| DoctorG | Free · context cost only | — |
| PPTX | Free · context cost only | — |
| Everything above | $0 of servers plus $0 of skills, the same in a quiet month and a busy one | $0/mo |
Subscriptions at their monthly plan price; one-off licences spread over 12 months.
Install only these and the bill is nothing: 5 servers and 5 skills, and all 17 tools.
Nothing in this stack carries a call allowance, so nothing here can run out before the month does. The bill is decided when you install it, not when you use it.
The spend cap is one figure for the whole account, set in your dashboard and checked when an order is placed. On the call path what stops a retry loop is the plan's call allowance and its rate limit, and a refused call is never billed.
The steps below come before the install lines: what to decide and scope before connecting anything, in the order it matters. The lines themselves follow, part by part, for the client you pick.
npx mcprush@latest stack add research-desk writes all 5 servers into Claude Code’s config in one go — or run the lines below one by one. The 5 skills are not included — add them with lines 06–10.
TAVILY_API_KEY — Your Tavily API keyPERPLEXITY_API_KEY — Your Perplexity API keyGOOGLE_DRIVE_OAUTH_CREDENTIALS — Path to a Google OAuth 2.0 Desktop app credentials JSON fileSet them where you start Claude Code — export TAVILY_API_KEY=… in your shell profile on macOS and Linux, setx TAVILY_API_KEY "…" on Windows, then open a new terminal. Claude Code reads them at start; the config keeps only the names.
# 01 · Tavily Search
claude mcp add --scope user --env 'TAVILY_API_KEY=${TAVILY_API_KEY}' --transport stdio tavily-search-mcp '--' npx -y mcp-tavily-search
# 02 · Perplexity Ai
claude mcp add --scope user --env 'PERPLEXITY_API_KEY=${PERPLEXITY_API_KEY}' --transport stdio perplexity-ai-mcp '--' npx -y @perplexity-ai/mcp-server
# 03 · Google Drive
claude mcp add --scope user --env 'GOOGLE_DRIVE_OAUTH_CREDENTIALS=${GOOGLE_DRIVE_OAUTH_CREDENTIALS}' --transport stdio piotr-agier-google-drive-mcp '--' npx -y @piotr-agier/google-drive-mcp
# 04 · Youtube Transcript
claude mcp add --scope user youtube-transcript-mcp '--' npx -y @kimtaeyoon83/mcp-server-youtube-transcript
# 05 · Csv Explorer
claude mcp add --scope user csv-explorer-mcp '--' npx -y @suppleaardvark/csv-explorer-mcp
# 06 · Paper Search Pro
npx mcprush@latest skill add o0000-code/paper-search-pro
# 07 · Crawl4AI
npx mcprush@latest skill add brettdavies/crawl4ai
# 08 · VideoDB
npx mcprush@latest skill add video-db/videodb
# 09 · DoctorG
npx mcprush@latest skill add glebis/doctorg
# 10 · PPTX
npx mcprush@latest skill add k-dense-ai/pptxnpx).Set-ExecutionPolicy -Scope CurrentUser RemoteSigned once, or type npx.cmd instead of npx (codex.cmd, gemini.cmd, grok.cmd likewise). PowerShell 7 needs neither.Nothing is connected until you approve it in Claude Code. mcprush is not in the path and meters nothing here.
Installed together, these tool surfaces add up. It is the first thing a security reviewer asks for, so what has been counted — and what nobody has counted yet — is on the page rather than in a PDF.
| Member | Tool surface | Write tools |
|---|---|---|
| Google Drive | not imported | not recorded |
| Tavily Search | not imported | not recorded |
A stack's blast radius is the union of its members, not the worst of them. That union cannot be taken here, because 2 members of 5 have no imported tool surface — so the figure a review asks for is missing rather than low, and a member marked not imported is one nobody has counted rather than one that cannot write.
1 skill in this stack instructs the agent to write — Paper Search Pro. A skill holds no permission of its own; it can only ask a server that already has the scope, and the approval gate still applies.
A stack is a default, not a verdict. These are the substitutions the maintainer would make, and what each one costs or saves.
Where the requirement is plain Google results — web, news, images, places — rather than search shaped for a model, Serper is the cheaper primitive.
Written by the maintainer, kept on the page rather than in a support thread.
It cannot give medical advice. The health member grades evidence and cites its sources; reading a study is not the same as being told what to do.
Scraping is bounded by what a site allows. A paywalled source stays paywalled, and the desk should say so rather than paraphrase a summary.
It has no opinion of its own about quality. Which of two contradicting sources is right is the analyst's judgement — the desk's job is to show both.
A stack has an owner: whoever keeps it re-checks the combination when a member changes, and the members themselves are published by the people named on each row.
Researches the account, writes the sequence, keeps the CRM current, and books the meeting without four tabs.
Gets the data out, trains or fine-tunes the model, and evaluates it before anybody calls it an improvement.
Turns the ticket, the thread and the source into a page somebody can follow, in the words your product already uses.