From ai-dev
Core Value-filtered competitive analysis: research user pain points, identify gaps within Core Value scope, file max 3 high-impact issues
How this skill is triggered — by the user, by Claude, or both
Slash command
/ai-dev:competitive-audit [focus-area][focus-area]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Evaluate project completeness, research competitors, analyze advantages/gaps, investigate improvement opportunities, and file actionable GitHub issues.
Evaluate project completeness, research competitors, analyze advantages/gaps, investigate improvement opportunities, and file actionable GitHub issues.
This is a long-running skill. Use TaskCreate to track phases and persist progress across /compact.
/competitive-audit or /competitive-audit [focus-area]$ARGUMENTS is provided, Phase 4 research focuses on that specific areaPhase 0: Core Value Check (GATE — stop if undefined)
Phase 1: Project Audit (assess current completeness)
Phase 2: Competitive Research (user pain points & competitor survey)
Phase 3: Gap Analysis (Core Value-filtered advantages & gaps)
Phase 4: Improvement Research (Core Value-scoped improvements only)
Phase 5: Issue Filing (max 3 issues + Won't Do recording)
Goal: Ensure the project has defined Core Values before proceeding.
CLAUDE.md and look for ## Core Values section## Core Values
1. {What is the ONE thing this product does better than anything else?}
2. {What is the second most important thing?}
3. {Optional: third}
Core Values are the single filter for this entire audit. Every finding, gap, and issue must pass through: "Does this directly strengthen a Core Value (one step, no indirect reasoning)?"
Goal: Accurately understand the current state of the project.
Use Agent (subagent_type: Explore) to investigate:
Create TaskUpdate with completeness score (1-10) and key findings:
Completeness: X/10
Core Values: [extracted from CLAUDE.md]
Strengths: [list]
Weaknesses: [list]
Unimplemented: [list]
Complexity: [package count, module count, config surface area]
Goal: Thoroughly research direct competitors, indirect competitors, and OSS alternatives.
Use Agent (subagent_type: general-purpose) to WebSearch in parallel:
This is the primary input for issue filing. Not competitor feature lists.
Compare UI/UX quality between this project and competitors:
If Playwright MCP is available and competitors are web apps:
Otherwise:
For each competitor, evaluate:
Analyze app store reviews for UI/UX insights (skip if not a mobile/desktop app):
"{app name}" app store review, "{app name}" user feedbackGoal: Clarify competitive positioning.
Cross-reference Phase 1 & 2 results. List clear advantages over competitors:
List areas where competitors are stronger, classified by Core Value relevance:
Core Value Related (actionable):
Non-Core Value (record but do NOT file issues):
For each disadvantage in 3b "Core Value Related", apply the one-step test:
"Does fixing this DIRECTLY strengthen a Core Value, without intermediate reasoning?"
Only ✅ Direct items proceed to Phase 4 and 5.
Create a 2-axis positioning map (ASCII art) to identify white space:
Axis A Axis B
┌──────────────────────────────────┐
Category 1 │ [Competitor A] │ │
│ [Competitor B] │ ★ Us │
└──────────────────────────────────┘
Goal: Research concrete improvement opportunities within Core Value scope only.
If $ARGUMENTS is provided, focus on that area (still filtered by Core Values).
If not specified, focus on the biggest Core Value-related gap from Phase 3.
Scope constraint: Only research improvements that passed the Phase 3c one-step test.
Use Agent (subagent_type: general-purpose) with extensive WebSearch:
Cover the following, all scoped to Core Value improvements:
Better models / libraries / tools
Optimization techniques
Architecture improvements
UX deepening opportunities
Latest research & academic papers
mcp__gemini-deepsearch__deep_search for thorough literature searchPrior implementations
Emerging technologies (no existing examples required)
mcp__gemini-deepsearch__deep_search and/or mcp__perplexity__perplexity_research to find:
For each improvement idea:
Goal: File research results as GitHub issues. Maximum 3 issues per audit run.
From Phase 3 and Phase 4 results, select at most 3 issues to file.
Selection criteria (all must be Yes):
If more than 3 candidates pass, rank by:
Rejected candidates: Record in the Won't Do section of the final report with reasoning.
Check existing labels with gh label list. Create missing labels as needed.
Priority labels:
## Won't Do)Each issue follows this structure:
## Summary
{1-2 line overview}
## Core Value Alignment
{Which Core Value this strengthens and how (one step, direct)}
## Motivation
{Why this improvement is needed, with USER PAIN POINT evidence (not just "competitor has it")}
## Complexity Cost
{New dependencies, config surface area, maintenance burden, affected existing features}
## Tasks
- [ ] {Concrete task 1}
- [ ] {Concrete task 2}
...
## References
- {Related URLs, papers, GitHub repos}
gh issue create to file#number in bodyresearch label to items requiring investigationFor items that were considered but rejected, append to the project's CLAUDE.md under ## Won't Do:
- **{Feature/idea}**: {Why not — e.g., "outside Core Value scope", "complexity cost too high", "indirect benefit only"}
This prevents future audits from re-proposing the same items.
Summary table of filed issues:
| # | Title | Priority | Core Value | Complexity Cost |
|---|-------|----------|------------|-----------------|
| #N | ... | P0 | ... | Low/Med/High |
Won't Do items added: [count]
After all phases, present to user:
gh command fails, output issue content as text for manual creationnpx claudepluginhub riox432/ai-dev-templates --plugin ai-devGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.