idea-memory

Scout research skill — Idea Memory

SKILL.md

GitHub

Skill: Idea Memory & Anti-Repetition

Track everything. Never repeat yourself. Learn from patterns.

Storage

Use Supabase table: scout_ideas (schema below)

Before Every Research Cycle

  1. Query scout_ideas for last 30 days
  2. Extract: sectors covered in last 7 days, product types sent in last 10 days
  3. HARD RULES:
    • No same sector within 5 days
    • No same product type within 10 days
    • No idea that's substantially similar to anything sent in the last 30 days
  4. Pick research sectors that are DUE for rotation

After Every Report

Insert a record:

  • date: today
  • status: 'sent' or 'killed'
  • title: one-line idea summary
  • sector: from the sector list below
  • product_type: from the product type list below
  • confidence_score: overall score (null if killed)
  • demand_score: demand validation score
  • revenue_confidence: high/medium/low
  • distribution_difficulty: easy/medium/hard
  • founder_fit: strong/moderate/weak
  • kill_reason: why it died (null if sent)
  • summary: 2-3 sentence description
  • build_time_estimate: estimated build time
  • pricing_model: revenue model
  • primary_channel: main distribution channel

Sector Tags

devtools, fintech, healthtech, edtech, proptech, creator-economy, smb-tools, enterprise, consumer, marketplace, ecommerce, vertical-saas, api-infrastructure, ai-ml-tools, legaltech, hrtech, climatetech, logistics, foodtech, sportstech, media-content, cybersecurity, insurtech, govtech, nonprofit-tools, construction-tech, automotive, travel-hospitality, agriculture, gaming

Product Type Tags

dashboard, workflow-automation, marketplace, api-wrapper, content-tool, analytics-platform, communication-tool, booking-scheduling, crm, internal-tool, monitoring-alerting, ai-assistant, browser-extension, data-pipeline, calculator-planner, directory-listing, template-library, integration-middleware, billing-invoicing, feedback-survey, social-tool, comparison-tool

Monthly Self-Assessment (1st of each month)

Send a Telegram summary:

  • Total ideas sent vs killed this month
  • Sectors covered (and any gaps)
  • Average confidence score
  • Top 3 ideas by confidence score
  • Most common kill reasons (patterns to watch)
  • Source quality: which research sources produced the best ideas
  • Recommendation: any adjustments needed to the research strategy