add-to-knowledge-base
Add new text resources to a knowledge base.
LABSCO SUMMARY
One key, and the breadth is the whole proposition.
Search, scraping, document conversion, media extraction, image generation and a code sandbox normally mean five accounts and five configurations; here they are one credential, which is worth something regardless of what you make of any individual tool. The extraction family is the most interesting part, because the same prompt-plus-JSON-mode pattern applies across documents, images, audio and video — so "pull these fields out of this" has one shape whatever the thing is. run-js-code and run-python-code are the pair to look at before installing: they execute code in a sandbox with a memory and timeout setting, which is a capability to enable deliberately.
What it is
A multi-capability server over the Dumpling AI API covering search, web extraction, document and media processing, generation, a knowledge base and sandboxed code execution.
What you get
- Google search with country, language, location and date range, optionally scraping the results, plus autocomplete suggestions and news search
- Maps and places: Google Maps by query or place ID, place search with business detail, and Google reviews with sorting
- Web extraction: a page scraped with JavaScript rendering and cleaning, a site crawled recursively to a depth and a limit, and structured data extracted against a schema
- Screenshots with full-page capture, a clip rectangle, device scale factor, cookie-banner blocking and auto-scroll
- Documents: any supported format converted to text with page selection, files converted to PDF, PDFs merged, and PDF metadata read and written
- Prompt-driven extraction from documents, images, audio and video, each with a JSON mode
- YouTube transcripts with timestamps and a language preference, and video trimmed between two timestamps
- Generation: agent completions with a thread ID, and AI images from a prompt and a model
- A knowledge base searched by query, and new text resources added to it
- JavaScript and Python executed in a sandbox with a timeout, a memory setting and optionally saved output files
Requirements
A Dumpling AI API key in DUMPLING_API_KEY.
Setup effort
One command plus a key — env DUMPLING_API_KEY=<YOUR_DUMPLING_API_KEY> npx -y mcp-server-dumplingai
USE CASES
01
Turning a competitor's site into something you can ask questions of
Search for the pages, crawl the site to a depth and limit, and put what comes back into a knowledge base so the next question is answered from stored text rather than another crawl.
02
Getting numbers out of a stack of documents
Convert what is unreadable, extract against a prompt with JSON mode when the fields are known, then compute over the result in the sandbox instead of asking the model to do arithmetic.
03
Researching a place before you go there
Maps and places searches return the business records; the reviews tool returns what people said about them, sorted and filtered by keyword.
04
Working from a video someone sent you
A transcript answers most questions about a talk without watching it; trimming and prompt-driven extraction handle the parts where the picture is the content.
05
Assembling a PDF pack for a client
Source files become PDFs, the PDFs become one document with PDF/A or PDF/UA conformance where it is required, and the metadata is written rather than left as whatever the converter guessed.
06
Checking what a page really renders for a visitor
A screenshot with JS rendering and cookie banners blocked shows what a person sees; the scrape of the same URL shows what a reader gets in text. Where they disagree is usually the finding.
27 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
convert-to-pdf
Convert various file formats to PDF.
crawl
Recursively crawl websites and extract content.
doc-to-text
Convert various document formats to plain text.
extract
Extract structured data from web pages using AI-powered instructions.
extract-audio
Extract structured data from audio files based on a prompt.
extract-document
Extract structured data from documents based on a prompt.
extract-image
Extract structured data from images based on a prompt.
extract-video
Extract structured data from videos based on a prompt.
generate-agent-completion
Generate AI text completions with customizable parameters.
generate-ai-image
Generate AI images from text prompts.
get-autocomplete
Retrieve Google autocomplete suggestions for a query.
get-google-reviews
Retrieve Google reviews for businesses or places.
get-youtube-transcript
Extract transcripts from YouTube videos with optional parameters for timestamps and language preferences.
merge-pdfs
Combine multiple PDF files into a single document.
read-pdf-metadata
Extract metadata from PDF files.
run-js-code
Execute JavaScript code in a sandbox.
run-python-code
Execute Python code in a sandbox.
scrape
Extract and parse content from any web page.
screenshot
Capture screenshots of web pages with customizable settings.
search
Perform Google web searches with customizable parameters.
search-knowledge-base
Search a knowledge base for relevant information.
search-maps
Search Google Maps for locations and businesses.
search-news
Search for news articles across multiple sources.
search-places
Search for places with detailed business information.
trim-video
Trim videos to a specific duration.
write-pdf-metadata
Write metadata to PDF files.
BEFORE YOU INSTALL
What to know first
Tools you get
27 tools when we connected. Generate-image is documented but not registered; generate-ai-image is available.
Requires
- npx on PATH
- DUMPLING_API_KEY in your environment
CONTEXT COST
Moderate Standing cost
Rated by the size of the 27 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
TERMINAL
env DUMPLING_API_KEY=<YOUR_DUMPLING_API_KEY> npx -y mcp-server-dumplingaiMCP Clients
Claude Desktop — Smithery
CLAUDE DESKTOP — SMITHERY · TERMINAL
npx -y @smithery/cli install @Dumpling-AI/mcp-server-dumplingai --client claudeCursor
CURSOR · JSON
{
"mcpServers": {
"dumplingai": {
"command": "npx",
"args": [
"-y",
"mcp-server-dumplingai"
],
"env": {
"DUMPLING_API_KEY": "<YOUR_DUMPLING_API_KEY>"
}
}
}
}
