tools/context_discovery.py

import re
import os
from typing import List
from app.tool.base import BaseTool
from app.utils.signal_registry import signal_registry
from app.tool.rss_matrix_sync import RSSMatrixSync
from app.logger import logger

class ContextDiscovery(BaseTool):
    """
    Scans the project/session context for 'Signals' (URLs, RSS feeds, API endpoints).
    Auto-configures and initializes RSS-CPs to populate the Signal Mesh.
    """
    name: str = "context_discovery"
    description: str = "Scan project files and environment for data signals to auto-initialize antennae."
    parameters: dict = {
        "type": "object",
        "properties": {
            "search_depth": {
                "type": "integer",
                "default": 1,
                "description": "How deep to scan the local project structure for signal sources."
            }
        }
    }

    async def execute(self, search_depth: int = 1) -> str:
        signals_found = []
        # 1. Scan .env and README for URLs
        files_to_scan = [".env", "README.md", "GEMINI.md", "package.json", "requirements.txt"]
        url_pattern = re.compile(r'https?://[^\s<>"]+|www\.[^\s<>"]+')

        for filename in files_to_scan:
            if os.path.exists(filename):
                try:
                    with open(filename, 'r') as f:
                        content = f.read()
                        urls = url_pattern.findall(content)
                        for url in urls:
                            # Heuristic: Prefer RSS/Atom/XML URLs
                            if any(ext in url.lower() for ext in ["rss", "atom", "xml", "feed"]):
                                signals_found.append(url)
                except Exception:
                    continue

        if not signals_found:
            # Fallback: Suggest common signals based on project keywords
            signals_found.append("https://news.ycombinator.com/rss") # Global tech signal

        # 2. Engage/Initialize the RSS-CPs
        sync_tool = RSSMatrixSync()
        engaged = []
        for url in list(set(signals_found)):
            logger.info(f"📡 Auto-tuning to discovered signal: {url}")
            result = await sync_tool.execute(url=url)
            engaged.append(f"{url}: {result[:50]}...")
            # Also register the raw feed as a broadcast
            signal_registry.broadcast(name=f"feed_{url}", source_type="rss_cp", data=url)

        report = f"Context Discovery complete. Engaged {len(engaged)} Signal Providers:\n" + "\n".join(engaged)
        return report