deal-sourcing-signals

Provide the company name and optionally its website, stage filter, and geography. The Skill runs six targeted web searches, classifies up to five signals per type with strength and sentiment, then Python scripts compute an overall deal score and readiness rating.

The scan covers six canonical signal types: HIRING, FUNDING, PRODUCT, TEAM, MARKET, and TECH. Each type is scored 0-100 and weighted, with hiring and funding weighted highest at 0.25 each.

Each signal's strength (1-10) is multiplied by a sentiment factor, aggregated per type with exponential normalization to 0-100, then weighted into an overall score. Scores of 70 or above, or a funding score of 60 or above, return MOVE FAST; 45 or above returns ENGAGE; otherwise MONITOR.

It requires Python 3 but uses only the standard library (json, math, os, sys, collections), so no pip installs are needed. Live signal data comes from Claude's web search, not from an external API.

The scorer expects raw_signals.json in the skill's output directory, written during the classification step. Run the web search and signal extraction steps first so the file exists before executing signal_scorer.py.

Signals depend on publicly indexed information, so stealth companies or unannounced rounds may score low despite real activity. Strength scoring is manual and subjective, and old or unverified news should be rated as weak or noise.