Recognize and avoid 15 common AI fluency anti-patterns that undermine effectiveness: over-reliance, prompt magic thinking, verification theater, and more.
Identifies 15 common AI usage mistakes that undermine effectiveness and provides diagnostic tools for remediation.
/plugin marketplace add leobessa/claude-plugins-ai-fluency/plugin install leobessa-ai-fluency@leobessa/claude-plugins-ai-fluencyThis skill inherits all available tools. When active, it can use any tool Claude has access to.
AI Fluency Anti-patterns catalogs recurring mistakes that undermine effective AI use. These patterns appear across all skill levels and persist because they "feel" productive while actually degrading outcomes.
Core Principle: Knowing what not to do is as important as knowing what to do.
Usage: Reference this when diagnosing AI effectiveness problems or training others.
What it is: Reduced vigilance because "AI is handling it."
Symptoms:
Root cause: Cognitive offloading without maintaining oversight.
Fix: Scheduled verification regardless of past performance. Never assume consistency.
What it is: Treating AI output as more authoritative than warranted.
Symptoms:
Root cause: Conflating fluency of expression with accuracy of content.
Fix: Treat all AI output as hypothesis. Verify against independent sources.
What it is: Continuing to iterate on bad prompts because of effort invested.
Symptoms:
Root cause: Emotional attachment to effort, not outcomes.
Fix: Set iteration limits. After 3 failed attempts, reframe the problem entirely.
What it is: Asking AI to do things without specifying what you actually want.
Symptoms:
Root cause: Unclear thinking about objectives, masked by AI's willingness to respond.
Fix: Define success criteria before prompting. If you can't specify what you want, think more before delegating.
What it is: Believing there's a secret prompt formula that unlocks perfect results.
Symptoms:
Root cause: Treating prompts as incantations rather than specifications.
Fix: Focus on clarity of requirements, not magic phrases. Understand what you're asking for.
What it is: Repeating prompts hoping for different results without changing approach.
Symptoms:
Root cause: Not treating AI interaction as a diagnostic process.
Fix: Each iteration should test a specific hypothesis. Document what you learn.
What it is: Providing massive context without curation, overwhelming the AI.
Symptoms:
Root cause: Avoiding the work of identifying what matters.
Fix: Curate context. Provide what's relevant, structured clearly. Less is often more.
What it is: Going through verification motions without actually verifying.
Symptoms:
Root cause: Wanting credit for diligence without the effort.
Fix: Define specific verification steps. Document what was checked.
What it is: Using AI's first response without iteration or verification.
Symptoms:
Root cause: Optimizing for speed over quality.
Fix: Budget time for review and iteration. First drafts are starting points.
What it is: Being more confident in AI accuracy than evidence supports.
Symptoms:
Root cause: Availability bias from successful interactions.
Fix: Track actual accuracy. Maintain skepticism regardless of history.
What it is: Trying to accomplish complex multi-step tasks in one prompt.
Symptoms:
Root cause: Not decomposing problems appropriately.
Fix: Break complex tasks into steps. Chain simpler prompts.
What it is: Expecting AI to reason well without structure.
Symptoms:
Root cause: Assuming AI will structure its own reasoning optimally.
Fix: Provide explicit reasoning frameworks. Tell AI how to think, not just what to think about.
What it is: Not establishing perspective or expertise for AI to adopt.
Symptoms:
Root cause: Leaving AI to guess what perspective to take.
Fix: Establish clear roles with relevant expertise and perspective.
What it is: Applying AI fluency practices inconsistently.
Symptoms:
Root cause: Not internalizing practices as habits.
Fix: Apply practices consistently. Quality should not vary by context.
What it is: Stopping improvement at "good enough."
Symptoms:
Root cause: Satisficing rather than optimizing.
Fix: Schedule regular practice and experimentation. Track improvement metrics.
When AI isn't working well, check:
ANTI-PATTERN DIAGNOSTIC
Cognitive:
□ Am I complacent about verification?
□ Am I treating AI as too authoritative?
□ Am I over-invested in this approach?
Process:
□ Did I clearly specify what I want?
□ Am I hoping for magic instead of engineering?
□ Am I iterating systematically?
□ Did I dump context without curation?
Verification:
□ Am I actually verifying or just skimming?
□ Did I accept first draft without review?
□ Am I overconfident about accuracy?
Structural:
□ Is this task too complex for one prompt?
□ Did I provide reasoning structure?
□ Did I establish an appropriate role?
Organizational:
□ Am I applying practices consistently?
□ Have I stopped improving?
Identified anti-patterns:
1. [Pattern]
2. [Pattern]
Remediation:
1. [Action]
2. [Action]
| Severity | Impact | Examples |
|---|---|---|
| Critical | Produces wrong outcomes | Output Authority Bias, Verification Theater |
| High | Significant quality loss | Vague Intent, First-Draft Acceptance |
| Medium | Reduced efficiency | Sunk Cost Prompting, Context Dumping |
| Low | Suboptimal results | Role-Free Prompting, Learning Plateau |
Anti-Pattern Awareness Complete When:
This skill should be used when the user asks about libraries, frameworks, API references, or needs code examples. Activates for setup questions, code generation involving libraries, or mentions of specific frameworks like React, Vue, Next.js, Prisma, Supabase, etc.
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.