/aaaEvaluate and improve ideas, product features, architecture, and code using evidence, industry best practices, research, gap analysis, and 3–4 world-class options.
Use /aaa when you want to know whether an idea, feature, architecture, workflow, strategy, or piece of code is excellent — or how to raise it to the highest practical quality level. It provides rigorous product, UX, technical, architecture, or code review backed by research and direct judgment.
Trigger phrases (from SKILL.md):
"is this excellent?""is this world-class?""AAA quality""worth building?""how do I improve this?""rigorous review of this feature / architecture / code"Invoke directly as a slash command, optionally followed by a description of what to evaluate:
/aaa [description or paste of the idea / feature / code / architecture to evaluate]The skill accepts any artifact: a short idea, a feature spec, a code snippet, an architecture diagram description, or a full file. If additional context (requirements, stack, competitors) is available, include it — the skill will use it.
Input:
/aaa We want to build an AI-powered note-taking app that auto-links related notes and surfaces forgotten ideas at the right moment. Is this AAA quality?Output / Result:
Initial classification: Promising but underdeveloped
The core idea (auto-linking + surfacing) is validated by tools like Roam, Obsidian, and Mem.
Differentiation is low without a clear edge on timing or context.
Gap analysis:
- Problem intensity: high (note retrieval is a known pain)
- Differentiation: weak — Mem.ai already does this with LLMs
- Feasibility: medium (embedding-based retrieval is mature)
3 world-class options:
1. Lean/elegant — Obsidian plugin, no new app, ship fast
2. AI-native — real-time bidirectional linking with semantic drift detection
3. Enterprise — team knowledge graph with access controls
Recommendation: Option 1 to validate, Option 2 if differentiation from Mem is confirmed.Input:
/aaa
```python
def get_user(id):
result = db.execute(f"SELECT * FROM users WHERE id = {id}")
return result.fetchone()
```
Is this world-class code?Output / Result:
Classification: Weak
Blockers:
- SQL injection: f-string interpolation directly into a query is a critical security vulnerability.
Fix: use parameterized queries — db.execute("SELECT * FROM users WHERE id = ?", (id,))
- No error handling: db failures silently return None
Improvements:
- Type hints: def get_user(id: int) -> Optional[User]
- Limit columns: SELECT id, name, email instead of SELECT *
- Use an ORM if available (SQLAlchemy, Django ORM)
Recommendation: Fix the injection blocker immediately; add type hints and column specificity next./aaa.