From Ambiguity to Specification.

Bridge the gap between product vision and technical execution with a standardized discovery workflow that reduces risk before you write a single line of code.

View on GitHub Read Documentation
[ASSUMPTION] ? Ambiguous Discovery Pack JTBD DDD ADR Schema Validation ✓ VALIDATED Spec-Ready
🎯 100% Schema Compliance
⚡ 35% Token Savings
📦 8 Methodologies
🌐 Multi-Agent Support

Why Discovery Pack Exists

Product teams using AI agents face a critical challenge: ambiguous requirements lead to wasted iterations, missed assumptions, and specifications that don't survive first contact with engineering. Discovery Pack solves this by transforming fuzzy ideas into rigorous, validated artifacts before you commit a single line of code.

Ambiguous Idea "Build a feature..." ??? Wasted Iterations Discovery Pack Spec-Ready Output Validated & Actionable

Key Features

Risk Mitigation

Proactively identify and test your riskiest assumptions before writing code. Move from "we think" to "we know."

Automated Governance

Ensure every specification meets your team's quality and compliance standards with built-in validation and logic checks.

Spec-Kit Native

Eliminate friction between product and engineering with a standardized handoff process that plugs directly into your workflow.

How It Works

1

Frame the Problem

Define users, jobs-to-be-done (JTBD), success metrics, and constraints using proven methodologies. Establish the "why" before jumping to solutions.

2

Analyze Options & Risks

Compare architectural approaches with scored trade-offs, auto-extract assumptions from your artifacts, and design validation experiments to test critical unknowns.

3

Generate Spec-Kit Ready Output

Produce schema-validated, implementation-ready artifacts that plug directly into GitHub Spec-Kit for seamless handoff from discovery to engineering.

Works With Your AI Agent

Platform-agnostic. Runs on any agent supporting the Agent Skills specification.

Claude Code
Anthropic
GitHub Copilot
CLI
VS Code
Agent Mode
Cursor
IDE

Choose Your Rigor Level

LITE MODE

Rapid Prototyping

15-30 min
  • 3 Core Artifacts
  • Problem Frame + Option Space + Handoff
  • Best for: POCs, small teams, low-risk features
Ideal for startups, internal tools, and exploratory work where speed matters more than exhaustive governance.
FULL MODE

Enterprise Grade

1-2 hours
  • 8 Complete Artifacts
  • Auto assumption extraction + validation plans
  • Best for: Production systems, compliance, high-risk
Required for production systems, regulated industries (finance, healthcare), and high-risk architecture with compliance requirements.

Detailed Comparison

Feature Lite Mode Full Mode
Time Required 15-30 minutes 1-2 hours
Artifacts Generated 3 outputs 8 outputs
Problem Frame (JTBD) ✓ ✓
NFR Constraints — ✓
Domain Model (DDD) — ✓
Option Space Analysis ✓ ✓
Auto Assumption Extraction — ✓
Validation Plans — ✓
Decision Logging (ADR) — ✓
Spec-Kit Handoff ✓ ✓
Schema Validation Manual 100% Auto
Best For POCs, Startups Enterprise, Compliance
Team Size < 5 people 5+ people

Powered by Intelligent Automation

35% Token Efficiency

Discovery Pack includes 7 Python scripts that automate repetitive validation, assumption extraction, and compliance checking. This automation layer reduces token consumption by 35% compared to pure AI-driven discovery, while ensuring 100% schema compliance and catching logical inconsistencies before they become problems.

  • → validate.py - Schema compliance
  • → extract_assumptions.py - Auto-generate risk register
  • → gate_detector.py - Logic gate detection
Epistemic Tagging Example:
## Requirements

[FACT] Current system handles 10K req/sec
[ASSUMPTION] Users expect <200ms response time
[HYPOTHESIS] Redis caching will reduce latency by 40%
[CONSTRAINT] Must comply with GDPR data retention
Automation Pipeline Tagged Artifacts Python Scripts Auto-Generated Assumption Register 04_assumptions-unknowns.md

Built For Product-Led Teams

Product Managers

Transform ambiguous stakeholder requests into structured requirements that engineering can validate. Move from "we should build X" to "here's why X solves user job Y with success metric Z."

Tech Leads & Architects

Design systems with unclear scope while maintaining architectural integrity and decision auditability. ADR logging ensures every "why" is captured for future maintainers.

Enterprise Teams

Meet compliance and governance requirements with automated validation, immutable decision logs (ADR), and epistemic tagging for assumption tracking. Built for regulated domains where auditability is non-negotiable.

Built on Proven Methodologies

JOBS-TO-BE-DONE (JTBD)
"People don't want a quarter-inch drill, they want a quarter-inch hole."
Clayton Christensen, JTBD Framework

Discovery Pack uses JTBD to ensure you're solving the right problem, not just building features. Frame requirements around user goals, not implementation details.

DOMAIN-DRIVEN DESIGN (DDD)
"The heart of software is its ability to solve domain-related problems for its user."
Eric Evans, Domain-Driven Design

Ubiquitous language and bounded contexts ensure your specification speaks the same language as your domain experts, preventing translation errors during implementation.

ARCHITECTURE DECISION RECORDS (ADR)
"Architecture decisions are design decisions that address architecturally significant requirements; they are perceived as hard to make and costly to change."
Michael Nygard, ADR Methodology

Immutable decision logs prevent revisiting old choices and provide a full audit trail for governance and compliance requirements.

Frequently Asked Questions

Do I need to install Python to use Discovery Pack?

No. The core workflow (SKILL.md + templates + schemas) works with any AI agent out-of-the-box. Python scripts are optional automation tools that provide 35% token savings and advanced validation, but are not required for basic usage.

How does Discovery Pack integrate with Spec-Kit?

Discovery Pack generates artifact 07_speckit-handoff.md which is directly consumable by GitHub Spec-Kit. Copy the constitution and specify sections from the handoff into your spec-kit workflow to continue from discovery to implementation planning.

Can I use Discovery Pack with GitHub Copilot or only Claude Code?

Discovery Pack is platform-agnostic. It works with Claude Code, GitHub Copilot CLI, VS Code Agent Mode, Cursor, and any agent that supports the Agent Skills specification or can execute local commands.

What's the difference between Lite and Full mode?

Lite mode (15-30 min) generates 3 core artifacts (Problem Frame, Option Space, Handoff) for rapid prototyping. Full mode (1-2 hours) generates all 8 artifacts with automated assumption extraction, validation plans, and ADR decision logging for enterprise-grade governance.

Is Discovery Pack suitable for regulated industries (healthcare, finance)?

Yes. Full mode includes JSON Schema validation (100% compliance), immutable decision logs (ADR), epistemic tagging for assumption tracking, and automated compliance checking — all critical for auditable, governed discovery processes.

Ready to Bridge the Gap?

Start your first discovery run in minutes. No installation required.

View on GitHub Read Installation Guide