discovery-sentinel

Discovery Sentinel

A Claude Code skill for analyzing product discovery and customer feedback session documents.

What it does

The Discovery Sentinel skill adopts the persona of a Principal Product Discovery Specialist for regulated B2B SaaS (ASX/ASIC context). It analyzes interview transcripts, meeting notes, and user research documents to produce:

• Feedback Classification Reports — per-signal classification with signal type, severity, bias flags, JTBD mapping, and routing recommendations
• Discovery Insight Briefs — synthesized atomic insights with confidence scores using the Gilad Confidence Meter
• Prioritised Opportunity Assessments — ranked opportunities scored with compliance-weighted RICE, WSJF, or MCDA frameworks

Key capabilities

• Jobs-to-be-Done (JTBD) extraction using 5 Whys decomposition
• Signal taxonomy classification (feature requests, UI friction, churn signals, workaround patterns, silence signals, etc.)
• Confidence scoring (0.1-10 scale) with explicit evidence requirements
• Cognitive bias detection and self-check protocols
• Anti-pattern flagging (discovery theatre, feature factory, HiPPO management)
• Regulatory awareness for ASX/ASIC compliance contexts
• Single document or batch analysis

Installation

/plugin install DiologIR/discovery-sentinel

Usage

Ask Claude to analyze discovery or feedback documents:

• "Analyze this interview transcript"
• "Review these customer feedback sessions and extract insights"
• "What are the key product signals from this user research?"
• "Prioritize the opportunities from these discovery notes"
• "Extract jobs-to-be-done from this customer interview"

Frameworks

Built on methodologies from:

• Teresa Torres — Continuous Discovery Habits / Opportunity Solution Trees
• Marty Cagan — Empowered Teams / Dual-Track Agile
• Itamar Gilad — Evidence-Guided / Confidence Meter
• Ryan Singer — Shape Up / Appetite-Driven Discovery
• Jobs-to-be-Done (JTBD) Switch Interviews

License

MIT