Files
leo-claude-mktplace/plugins/clarity-assist/skills/prompt-patterns/optimization-rules.md
lmiranda e5ca804692 feat: v3.0.0 architecture overhaul
- Rename marketplace to lm-claude-plugins
- Move MCP servers to root with symlinks
- Add 6 PR tools to Gitea MCP (list_pull_requests, get_pull_request,
  get_pr_diff, get_pr_comments, create_pr_review, add_pr_comment)
- Add clarity-assist plugin (prompt optimization with ND accommodations)
- Add git-flow plugin (workflow automation)
- Add pr-review plugin (multi-agent review with confidence scoring)
- Centralize configuration docs
- Update all documentation for v3.0.0

BREAKING CHANGE: MCP server paths changed, marketplace renamed

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-20 16:56:53 -05:00

135 lines
3.1 KiB
Markdown

# Prompt Optimization Rules
## Core Rules
### Rule 1: Specificity Over Generality
| Instead of | Use |
|------------|-----|
| "Make it better" | "Reduce load time to under 2 seconds" |
| "Add some validation" | "Validate email format and require 8+ char password" |
| "Handle errors" | "Show toast notification on API failure, log to console" |
### Rule 2: Include Context
Every good prompt includes:
- **What**: The action/feature/fix needed
- **Where**: Location in codebase or UI
- **Why**: Purpose or problem being solved
- **Constraints**: Technical limits, compatibility, standards
### Rule 3: Define Success
Specify how to know when the task is done:
- Acceptance criteria
- Test cases to pass
- Behavior to verify
### Rule 4: Scope Boundaries
Explicitly state:
- What IS in scope
- What is NOT in scope
- What MIGHT be in scope (user's call)
## Anti-Patterns to Detect
### Vague Requests
Triggers: "improve", "fix", "update", "change", "better", "faster", "cleaner"
Response: Ask for specific metrics or outcomes
### Scope Creep Signals
Triggers: "while you're at it", "also", "might as well", "and another thing"
Response: Acknowledge, then isolate: "I'll note that for after the main task"
### Assumption Gaps
Triggers: References to "the" thing (which thing?), "it" (what's it?), "there" (where?)
Response: Echo back specific understanding
### Conflicting Requirements
Triggers: "Simple but comprehensive", "Fast but thorough", "Minimal but complete"
Response: Prioritize: "Which matters more: simplicity or completeness?"
## Question Templates
### For Unclear Purpose
```
**What problem does this solve?**
1. [Specific problem A]
2. [Specific problem B]
3. Combination
4. Different problem: ____
```
### For Missing Scope
```
**What should this include?**
- [ ] Feature A
- [ ] Feature B
- [ ] Feature C
- [ ] Other: ____
```
### For Ambiguous Behavior
```
**When [trigger event], what should happen?**
1. [Behavior option A]
2. [Behavior option B]
3. Nothing (ignore)
4. Depends on: ____
```
### For Technical Decisions
```
**Implementation approach:**
1. [Approach A] - pros: X, cons: Y
2. [Approach B] - pros: X, cons: Y
3. Let me decide based on codebase
4. Need more info about: ____
```
## Optimization Checklist
Before proceeding with any task, verify:
- [ ] **Specific outcome** - Can measure success
- [ ] **Clear location** - Know where changes go
- [ ] **Defined scope** - Know what's in/out
- [ ] **Error handling** - Know what happens on failure
- [ ] **Edge cases** - Major scenarios covered
- [ ] **Dependencies** - Know what this affects/relies on
## ND-Friendly Adaptations
### Reduce Cognitive Load
- Maximum 4 options per question
- Always include "Other" escape hatch
- Provide examples, not just descriptions
### Support Working Memory
- Summarize frequently
- Reference earlier decisions explicitly
- Don't assume user remembers context
### Allow Processing Time
- Don't rapid-fire questions
- Validate answers before moving on
- Offer to revisit/change earlier answers
### Manage Overwhelm
- Offer to break into smaller sessions
- Prioritize must-haves vs nice-to-haves
- Provide "good enough for now" options