Community Research MCP

""" GitHub Issues & Discussions Search. Search GitHub's issue tracker sorted by community reactions to find the most relevant discussions and solutions. API: https://docs.github.com/en/rest/search Rate Limits: 10/min (anonymous), 30/min (authenticated) """ import logging import os from typing import Any, Optional import httpx __all__ = ["search"] # ══════════════════════════════════════════════════════════════════════════════ # Configuration # ══════════════════════════════════════════════════════════════════════════════ API_BASE = "https://api.github.com" API_TIMEOUT = 30.0 API_TOKEN = os.getenv("GITHUB_TOKEN") logger = logging.getLogger(__name__) # Words to filter from queries (too generic) STOPWORDS = frozenset( { "the", "a", "an", "and", "or", "but", "in", "on", "at", "to", "for", "of", "with", "by", "from", "how", "what", "where", "when", "why", "solution", "fix", "help", "error", "issue", "problem", "code", "example", } ) # ══════════════════════════════════════════════════════════════════════════════ # Search Function # ══════════════════════════════════════════════════════════════════════════════ async def search( query: str, language: Optional[str] = None, *, max_results: int = 30, ) -> list[dict[str, Any]]: """ Search GitHub issues and discussions. Args: query: Search query string language: Programming language filter max_results: Maximum results to return Returns: List of issues with title, url, state, comments, snippet Example: >>> results = await search("memory leak", language="python") """ # Build simplified query (GitHub has 256 char limit) search_query = _build_query(query, language) headers = {"Accept": "application/vnd.github.v3+json"} if API_TOKEN: headers["Authorization"] = f"token {API_TOKEN}" params = { "q": search_query, "sort": "best-match", # Changed from "reactions" to prioritize relevance "order": "desc", "per_page": min(max_results, 100), } try: async with httpx.AsyncClient(timeout=API_TIMEOUT) as client: response = await client.get( f"{API_BASE}/search/issues", params=params, headers=headers, ) response.raise_for_status() data = response.json() return [ { "title": item.get("title", ""), "url": item.get("html_url", ""), "state": item.get("state", ""), "comments": item.get("comments", 0), "reactions": item.get("reactions", {}).get("total_count", 0), "snippet": (item.get("body") or "")[:1000], "source": "github", } for item in data.get("items", []) ] except httpx.HTTPStatusError as e: if e.response.status_code == 422: logger.warning(f"Query too complex: {query[:50]}...") else: logger.warning(f"HTTP {e.response.status_code}") return [] except Exception as e: logger.error(f"Search failed: {e}") return [] def _build_query(query: str, language: Optional[str], max_length: int = 256) -> str: """Build a simplified GitHub search query within length limits.""" suffix = f"is:issue{f' language:{language}' if language else ''}" available = max_length - len(suffix) - 1 # Extract important keywords words = [] for word in query.split()[:10]: clean = word.strip().lower() is_technical = ( len(clean) > 2 and clean not in STOPWORDS or any(c.isdigit() for c in clean) ) if is_technical: words.append(word) # Build query within limit result = "" for word in words: test = f"{result} {word}".strip() if len(test) <= available: result = test else: break return f"{result} {suffix}".strip() # ══════════════════════════════════════════════════════════════════════════════ # Backward Compatibility # ══════════════════════════════════════════════════════════════════════════════ search_github = search