What problem does it solve? Sales teams waste time on cold prospects because they lack visibility into which target companies are actively showing buying intent. This Skill automates the discovery and scoring of intent signals (funding, hiring, leadership changes, product launches) so outreach focuses on companies with real, recent, attributable triggers. ## Core Features & Use Cases - Multi-source signal collection: Runs Parallel web search (always on), optional Firecrawl website crawls (native API or pre-crawled MCP pages), and optional Parallel structured enrichment for funding, hiring, and tech stack fields. - Two-axis intent scoring: Scores each signal 1-100 using an attribution gate (direct/adjacent/generic) plus offering fit, with freshness gating, domain verification, and prompt-injection hardening baked into the script. - Flexible LLM backends: Default agent backend collects raw evidence for in-context scoring with no external LLM cost; claude-cli and legacy OpenRouter backends support fully autonomous runs. - Use Case: Given a CSV of ICP-scored companies, run the skill to produce signals.csv where each company has an overallScore, scored signals with source URLs and dates, and a one-line outreach hook. ## Quick Start Ask the agent to run signal search on your client directory, for example: run signal search on the acme-gtm folder with a 2-month lookback and a test limit of 5 companies.
