Purpose
Scout MCP is a business and market intelligence MCP server for AI agents, positioned as the "Google for AI agents." Unlike traditional search engines that return web pages, it returns clean, structured JSON that agents can directly reason over, for tasks such as researching companies, analyzing competitors, and tracking market trends.
Main Capabilities
Provides 6 intelligence tools:
- Company research (scout_company): returns fields such as name, domain, description, industry, founded, headquarters, employee_range, funding, tech_stack, social_profiles, recent_news, top_competitors, key_people, confidence, confidence_breakdown, data_quality_grade, data_freshness, sources_used, sources_failed, etc.
- Market research: returns fields such as market_name, query, depth, market_size, market_size_projections, cagr, key_players, trends, growth_drivers, risks, source_links, confidence, etc.
- Competitor analysis (scout_competitors): returns fields such as target, competitors (each item includes name, domain, positioning, pricing, strengths, weaknesses, key_differentiator), market_positioning_summary, etc.
- Trend tracking (scout_trends): returns fields such as topic, timeframe, sentiment (score, label), trending_direction, key_developments, related_topics, social_buzz, etc.
- Product intelligence: returns name, category, description, pricing, ratings, features, ideal_for, alternatives, recent_updates, etc.
- Person research: returns fields such as name, current_role, company, location, background_summary, social_profiles, recent_activity, notable_achievements, etc.; requires Pro tier or higher, and the free tier returns HTTP 403.
Data sources aggregate DuckDuckGo, NewsAPI, Wikipedia, web scraping, and social profiles. Every response includes data_quality_grade (data quality letter grade) and confidence_breakdown (per-source confidence details); different sources have different weights in the overall confidence calculation.
The competitor extraction engine uses 6 types of regex patterns and 350+ stop words, including VS patterns, comma/and separated lists, numbered/bulleted lists, heading patterns, context patterns (such as "like X" "such as X"), and title-case names; the multi-word validator removes article titles, verb-prefixed names, pronoun-prefixed names, role-suffixed names, suspected person names, large parent company names, and platform names.
How to Use
Integration methods include:
- Claude Desktop: add to ~/Library/Application Support/Claude/claude_desktop_config.json.
- Cursor: add in Settings > MCP.
- VS Code (Copilot MCP): add to .vscode/mcp.json.
- pip self-hosting: pip install scout-intel-mcp. The API key is passed via the X-Api-Key request header; without a key, the free tier is used (50 requests per day). Responses are cached in an in-memory dictionary for 24 hours, with cache statistics and clear endpoints provided; rate limit information is included in the _meta field of every response. The DuckDuckGo free API has rate limits, and Scout MCP handles them with adaptive exponential backoff, retrying once before each call fails; backoff status can be viewed at GET /api/health. Tech stack baseBased on FastMCP, httpx, BeautifulSoup4, Pydantic (current version v0.1.0). Environment variables and local development methods are described in the repository.
Use cases
Suitable for AI agents performing company due diligence, industry and market size research, competitor positioning and pricing comparison, topic trend and sentiment tracking, product selection comparison, public figure background research, and other tasks, especially suitable for agent workflows that require structured JSON output with confidence and quality ratings rather than web page text.
Limitations
Person research requires the Pro tier or higher and is unavailable on the free tier; the free tier allows 50 requests per day; planned future data sources include the Crunchbase API (funding data), SimilarWeb API (traffic data), GitHub API (developer tools), and SEMrush API (SEO data), which are not yet available. Specific pricing and rate limits for each tier, environment variable details, local development steps, and so on are subject to the repository documentation; pricing, price, and performance information not listed in this introduction is not yet available.
