AgentSkillsCN

assumption-validation

在做出承诺前测试假设是否成立。在假设确定性低且风险高时，请使用此技能。

SKILL.md
--- frontmatter
name: assumption-validation
description: Test whether assumptions are true before making commitments. Use when assumptions have low certainty and high risk.

Assumption Validation

Overview

Test whether an assumption is true or false before making commitments based on it.

When to Use

  • •When an assumption has low certainty and high risk
  • •Before major resource commitments
  • •When stakeholders challenge your reasoning
  • •During Empathize and Define phases

How to Apply

1. State the Assumption Clearly

Make it testable. Transform vague beliefs into specific claims:

  • •❌ "Users want better tools"
  • •✅ "Users spend >2 hours/week on manual data entry and would use an automated solution"

2. Define What Would Validate or Invalidate

Be explicit about criteria:

  • •Validates: 8+ out of 10 users report >2 hrs/week on manual entry
  • •Invalidates: <5 out of 10 users report this, or they prefer manual control
  • •Inconclusive: Mixed results, need different approach

3. Choose Validation Method

Match method to assumption type:

User behavior/needs: Interviews, observation, surveys Technical feasibility: Spikes, prototypes, vendor demos Market conditions: Market research, competitor analysis Business viability: Financial modeling, expert consultation

4. Execute Validation

Conduct research with focus on disproving, not confirming:

  • •Ask open-ended questions
  • •Observe actual behavior, not just stated preferences
  • •Look for contradictory evidence
  • •Talk to diverse user types

5. Update Status

Record findings in currentstate.json:

  • •Validated: Evidence supports the assumption
  • •Invalidated: Evidence contradicts it
  • •Partially validated: More complex than assumed
  • •Needs more research: Inconclusive

6. Act on Findings

If validated: Proceed with confidence, but stay alert for new evidence

If invalidated:

  • •Update problem framing
  • •Revise approach
  • •Generate new assumptions
  • •May need to pivot

If partially validated:

  • •Refine the assumption
  • •Identify what's true and what's not
  • •Adjust plans accordingly

Example

Assumption: "Field technicians need offline access"

Validation Plan:

  • •Interview 8 field technicians
  • •Ask about connectivity at work locations
  • •Observe their current workarounds
  • •Ask what happens when connection drops

Findings:

  • •7/8 work in areas with spotty connectivity
  • •All have experienced data loss from connection drops
  • •All use workarounds (paper notes, photos) when offline
  • •Strong preference for offline-first design

Result: VALIDATED — Offline access is a critical requirement

Action: Prioritize offline functionality in ideation phase

Tips

  • •Seek to disprove, not confirm
  • •Sample diverse users, not just friendly ones
  • •Observe behavior, don't just ask
  • •Document exact evidence, not interpretations
  • •Update currentstate.json immediately
  • •One assumption may spawn new assumptions