skills/vcskills/skills/deal-sourcing

deal-sourcing

Installation
$ npx skills add https://github.com/vcskills/deal-sourcing
Skill.md

Deal Sourcing

Systematically identify, track, and qualify investment opportunities across inbound, outbound, and referral channels.

Quick Start

Use this skill when:

  • You are building or refreshing a systematic sourcing engine from scratch
  • Inbound deal flow is unstructured and living across email, Slack, and spreadsheets
  • You want to score and prioritize opportunities before the first call
  • You are mapping a new vertical and need a target company list quickly
  • You need to report pipeline health metrics to the partnership

How It Works

Deal Sourcing connects to your existing communication stack — email, Slack, CRM — and extracts structured deal signals from unstructured inbound. Each opportunity is automatically normalized into a deal record with company name, stage, vertical, and source.

The skill then applies your thesis scoring rubric, assigning a fit score from 0–100 based on configurable criteria such as check size alignment, sector fit, founder background, and market size. Deals above your threshold are surfaced for review; those below are logged but deprioritized.

For outbound, the skill generates target lists by pulling from public databases, filtering by your parameters, and ranking by signal strength — recent funding rounds, hiring velocity, and founder social signals.

Thesis Configuration

Define your thesis as a JSON config with weighted criteria. For example: `{ sector: ['climate', 'fintech'], stage: ['seed', 'series-a'], checkSize: { min: 500000, max: 3000000 }, geoFocus: ['US', 'EU'] }`. The skill uses this config to score every inbound deal automatically.

Weights are adjustable — a sector match might be worth 40 points while geography is worth 10. This lets you tune sourcing sensitivity without rewriting logic.

CRM Integration

Deal Sourcing writes structured records to Affinity, Salesforce, or Notion automatically. Each record includes the raw source, normalized fields, thesis score, and a one-paragraph AI-generated summary. Existing records are updated rather than duplicated.