get_sentimentLABSCO SUMMARY
Divergence is the number you cannot get from an English-only feed.
A sentiment score on its own is a crowded product. What is not crowded is knowing that the English-language coverage of TSMC and the Traditional Chinese coverage disagree, and by how much — the README puts the threshold worth noticing at above 0.3 in absolute terms. The rest is well-judged plumbing: the score comes back with its confidence and the three articles driving it, so the answer is checkable rather than a number to trust. Watch the quota arithmetic on watchlists, since a batch bills per ticker.
What it is
An MCP server over Praesentire, a financial news sentiment API that reads both English and Traditional Chinese coverage. The distinctive tool is the comparison: the same ticker scored separately in each language, with a divergence number, which is a signal that only exists because both sides are read.
What you get
- An aggregated sentiment score from -1 to +1 for one ticker, with a confidence value, a bull and bear distribution, and the latest 3 articles with the rationale behind each —
get_sentiment - The same aggregates for up to 50 tickers in one call, for a watchlist rather than a name —
get_sentiment_batch - English against Traditional Chinese sentiment on the same ticker, side by side with a divergence score, aimed at cross-market signals —
compare_languages - Sources on both sides: Reuters, Bloomberg, Yahoo Finance and CNBC in English, Economic Daily News (經濟日報), Commercial Times (工商時報) and Anue (鉅亨) in Chinese, updated every 10 minutes
- Errors mapped to messages a model can act on rather than status codes — an invalid key says to create a new one, a rate limit says which tier was exceeded
Requirements
A Praesentire API key in PRAESENTIRE_API_KEY, in the form pr_.... The free tier is 100 requests a day with no credit card; paid tiers go from $19 to $499 a month. Each tool call is one request, and a batch of N tickers counts as N requests — the batch tool saves round trips, not quota. Launch with npx -y praesentire-mcp over stdio; the package is praesentire-mcp (0.1.0 in package.json). PRAESENTIRE_BASE_URL overrides the endpoint for self-hosting or staging.
Setup effort
One command plus a key — npx praesentire-mcp, then supply credentials
USE CASES
01
Understanding a move you did not expect
Pull the sentiment on the ticker and read the three articles with their rationales, rather than the score alone.
02
Spotting a cross-market disagreement
Compare the English and Traditional Chinese sentiment on the same name and read the divergence score.
03
Checking a whole watchlist at once
Send the tickers in one batch call, remembering that each one still counts against the daily quota.
04
Following Taiwan-listed names properly
Use the Chinese-language sources rather than relying on English wires that cover them thinly.
3 tools
Listed in project docs. Your client may show a different list.
get_sentiment_batchcompare_languagesBEFORE YOU INSTALL
What to know first
Tools you get
3 tools named in the project's documentation.
Requires
- npx on PATH
- PRAESENTIRE_API_KEY in your environment
Where it runs
All 3 documented clients run the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
npx praesentire-mcpMCP Clients
Claude Code, Claude Desktop, Cowork
JSON
{
"mcpServers": {
"praesentire": {
"command": "npx",
"args": [
"-y",
"praesentire-mcp"
],
"env": {
"PRAESENTIRE_API_KEY": "<YOUR_PRAESENTIRE_API_KEY>"
}
}
}
}
