assumption-challenger

I was intrigued by how assumption-challenger frames validation as a collaborative debugging tool rather than just error-catching—the scoring reflects solid execution, but I'm curious about the edge cases where assumptions hide in plain sight.

Links

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 81/100, solid B-grade territory. This is graded against Anthropic's PDA (Progressive Disclosure Architecture) best practices and the 5-pillar rubric. Your strongest area is Utility (17/20)—the framework actually solves real problems. Weakest is Writing Style (7/10) and PDA (22/30)—mostly because the main file is bloated and references a skill file that doesn't exist.

What's Working Well

• Problem-solving power is real: The categorization (Technical, Market, Organizational, Process, Resource) actually maps to how assumptions break in the real world. That's not accidental design.
• Templates are comprehensive: The verdict system (Valid/Questionable/Invalid/Unknown) and the 4-step process give people a clear path without hand-holding them to death.
• Trigger phrases are tight: "validating assumptions before launching", "reviewing strategic decisions"—these are specific enough that developers know exactly when to pull this skill in.
• Examples hit the mark: The real-world assumption examples (market size, technology viability, team capability) feel grounded, not theoretical.

The Big One

Your main SKILL.md file is 356 lines without a table of contents, and you're referencing wishful-thinking-patterns.md which doesn't exist. This kills your PDA score immediately because:

1. Long files = poor navigation = token waste
2. Broken references = users get stuck
3. Violates the "one reference level deep" pattern

Here's the fix: Split the file. Keep core process (When to Use, Why Assumptions Matter, the 4 steps, output format) in SKILL.md. Move the detailed categorization examples and challenge pattern library to a properly created challenge-patterns.md file. Update the reference. You'll gain ~3-4 points just from this.

Other Things Worth Fixing

1. Add a TOC to both files - SKILL.md and challenge-questions.md both exceed 100 lines. Three-line TOC each = better navigation and +1-2 points. Put it right after the metadata.

2. Condense the category examples - Each assumption category (Technical, Market, etc.) has redundant "Common patterns" sections. Cut to one example per category; link to challenge-questions.md for the full list. Saves ~40 lines and improves token efficiency.

3. Fix the voice in challenge templates - Lines 179-218 use second-person ("You assume", "What makes you different?"). Rewrite to imperative: "Challenge: Compare this assumption against industry data." More consistent, +1 point.

4. Verify or remove integration references - You mention antipattern-detector and validation-report-generator skills. If they don't exist, remove the integration diagram or add a "(if available)" disclaimer.

Quick Wins

• Create the missing challenge-patterns.md file and move ~100 lines there (+2 points)
• Add TOCs to both files (+1 point)
• Fix broken reference or create file (+2 points)
• Condense examples by 30% (+1 point)

That's +6 points with an afternoon of work—puts you at 87/100, solid A territory.

Checkout your skill here: SkillzWave.ai | SpillWave We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.