tk-discovery

What problem does it solve?

tk-discovery converts stakeholder interviews and mixed source documents into a high-fidelity, decision-preserving Discovery Brief plus project configuration that can be used as a single source of truth for downstream planning and implementation.

Core Features & Use Cases

• Guided end-to-end discovery pipeline: runs a structured, multi-phase flow from mode detection and intake to freeze-map creation, tiered feature deep-dives, and final brief synthesis.
• Anti-drift decision preservation: enforces non-negotiable rules to prevent silent scope/stakeholder changes, hallucinated facts, and lossy merging of sources.
• Confidence tagging & drift detection: applies confidence semantics (confirmed vs inferred vs assumptions vs open questions) and includes audit artifacts to surface what came from where.
• Gap-driven stakeholder clarification: asks targeted questions only for genuine open questions, using checkpoint pauses to ensure explicit user approvals.
• Tiered feature specification: categorizes features into Configure (S), Extend (M), and Build custom (L) and routes Build-custom work through interactive clarifications with ADR flagging.

Quick Start

Run the workflow via the slash command: ask the AI to execute /discovery with your inputs and confirm any open questions presented during the checkpoints.