knowledgeLABSCO SUMMARY
Paid retrieval with the price named in the call.
Every search carries a maximum price per thousand queries as a required argument, which makes cost a parameter rather than a surprise on the invoice. Choosing between proprietary indexes and the web per query is what separates it from a general web search tool.
What it is
A Valyu knowledge retrieval server. A search runs over proprietary sources, the web, or both, with a maximum price per thousand queries as a required argument, and optional control over which indexes are searched, how many results survive reranking, the similarity floor, and whether the query is rewritten first. A second tool submits feedback against a transaction.
What you get
- Retrieval over proprietary sources, the web, or both, chosen per call
- A maximum price per thousand queries required on every search, so cost is bounded at the call site
- Specific indexes named per query, such as valyu/valyu-arxiv and valyu/valyu-wikipedia
- The number of results returned after reranking, and a minimum similarity score for what is included
- Optional query rewriting for better retrieval performance
- Feedback submitted against a transaction id with a sentiment rating
Requirements
A Valyu API key. The published route is the Docker image ghcr.io/tiovikram/valyu-mcp-server, run with the key passed in as an environment variable.
Setup effort
Follow the published steps — docker pull ghcr.io/tiovikram/valyu-mcp-server
USE CASES
Researching something the open web answers badly
Search proprietary indexes instead of, or alongside, the web — naming the ones that should be consulted, arXiv and Wikipedia among them.
Keeping retrieval inside a budget
Each search takes a maximum price per thousand queries, so an agent looping over many questions cannot quietly outspend what you intended.
Getting fewer, better results
Set how many results survive reranking and the minimum similarity score, so a long tail of weak matches never reaches the model's context.
Improving what comes back next time
Feedback goes back against the transaction id with a sentiment rating, tying the judgement to the specific retrieval it refers to.
2 tools
Listed in project docs. Your client may show a different list.
feedbackBEFORE YOU INSTALL
What to know first
Tools you get
2 tools named in the project's documentation.
Requires
VALYU_API_KEY in your environment
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
docker pull ghcr.io/tiovikram/valyu-mcp-server
docker run -i --rm -e VALYU_API_KEY=<YOUR_VALYU_API_KEY> ghcr.io/tiovikram/valyu-mcp-serverMCP Clients
Claude Desktop
{
"mcpServers": {
"valyu": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"-e",
"VALYU_API_KEY",
"ghcr.io/tiovikram/valyu-mcp-server"
],
"env": {
"VALYU_API_KEY": "<YOUR_VALYU_API_KEY>"
}
}
}
}