MCP server for scraping financial news from multiple sources including Reuters, MarketWatch, CNBC, Yahoo Finance, Financial Times, Seeking Alpha, and Investing.com
Financial News MCP has clear, well-structured tool definitions with proper schemas and descriptions. All 4 tools are explicitly registered with inputSchema, descriptions, and parameter constraints (enums, min/max). Naming follows verb_noun convention (get_*, search_*, list_). However, output schemas are not documented, responses are formatted as plain text rather than structured objects, limiting downstream tool chaining. Parameter descriptions are adequate but lack dependency hints and format guidance. No error handling guidance for invalid sources or network failures. No tool annotations (readOnlyHint, etc.) despite all tools being read-only.
Fetch the latest financial news headlines from major sources including Reuters, MarketWatch, CNBC, Yahoo Finance, Financial Times, Seeking Alpha, and Investing.com.
Fetch the latest news from a specific financial news source.
List all available financial news sources and their categories.
Search for financial news articles matching a keyword or phrase across multiple news sources.
Output schemas not documented. Tools return formatted text strings rather than structured objects with typed fields. LLMs cannot extract specific fields (title, URL, source, date) for downstream processing or chaining.
No tool annotations despite all tools being read-only. Missing readOnlyHint annotation prevents clients from optimizing caching and retry logic.
No error recovery guidance. If 'source' parameter is invalid or network fails, tool returns plain error text. LLM has no guidance on retry strategy, available alternatives, or next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 69 | 2026-07-28+ | v2 |
Parameter descriptions lack format guidance. 'query' in search_financial_news has no length limits, regex patterns, or examples of valid queries. LLMs may pass overly long or malformed strings.
No pagination support. Tools return up to 20 articles per call but do not support offset/cursor-based pagination. Large result sets cannot be retrieved incrementally, risking context window exhaustion.