ultrathink-trigger
Auto-escalate reasoning depth for complex problems. Use proactively when encountering architecture, debugging, or unfamiliar code.
安装与使用
请根据来源仓库和你的 AI 工具说明安装此技能。先阅读技能文件和依赖要求,再添加到工作流。
查看来源 ↗使用指南技能说明
Ultrathink Trigger
Automatically invoke maximum reasoning depth for complex problems.
Complexity Indicators
| Signal | Weight | Example |
|---|---|---|
| Files affected > 5 | +2 | Refactoring across modules |
| Unfamiliar language/framework | +3 | First time with codebase |
| Architecture decision | +4 | Choosing patterns |
| 2+ failed fix attempts | +3 | Root cause unclear |
| Multi-domain interaction | +2 | Frontend + Backend + DB |
| Security implications | +3 | Auth, permissions, secrets |
| Performance-critical code | +2 | Hot paths, algorithms |
| Breaking change potential | +3 | API changes, schema migrations |
Threshold
If total weight >= 5: Trigger ultrathink mode
Invocation
Prefix complex analysis with one of:
- "Think harder about this:"
- "ultrathink:"
- "Take your time to deeply analyze:"
These phrases trigger Claude's extended thinking mode for more thorough analysis.
Integration Points
With /fix command
After 2 failed attempts, automatically invoke:
ultrathink: What are all the possible root causes?
What assumptions am I making? What haven't I checked yet?
With /brainstorm command
Auto-triggers when topic complexity >= 5 (or --deep flag). Applied to EVALUATE and SYNTHESIZE phases:
ultrathink: Deeply analyze these options considering trade-offs,
edge cases, maintainability, and long-term implications.
Flags:
-
--deep- Force ultrathink regardless of complexity -
--quick- Suppress auto-detect, stay fast
With orchestrator INNOVATE phase
For architecture decisions:
ultrathink: Evaluate these approaches considering
maintainability, performance, security, and complexity.
With sequential-thinking
For multi-step complex analysis:
mcp__sequential_thinking__sequentialthinking({
thought: "ultrathink: [complex problem]",
thoughtNumber: 1,
totalThoughts: 10, // More steps for deep analysis
nextThoughtNeeded: true
})
When to Use Proactively
- Architecture Decisions: Before proposing patterns or structure
- Debugging Deadlock: After 2+ failed fix attempts
- Security Analysis: Any auth, permission, or secret handling
- Performance Optimization: Before optimizing hot paths
- Breaking Changes: Before proposing API/schema changes
- Unfamiliar Code: First encounter with a new codebase area
Example Workflow
1. Encounter complex problem
2. Calculate complexity weight:
- Multi-file change: +2
- Architecture decision: +4
- Total: 6 (>= 5, triggers ultrathink)
3. Invoke: "ultrathink: Design the data flow for..."
4. Extended analysis produces thorough solution
5. Proceed with implementation
Anti-Patterns
- Using ultrathink for simple, well-understood tasks
- Skipping ultrathink for complex problems to "save time"
- Not recalculating complexity when scope changes
使用示例
查看来源仓库中的使用示例与配置说明。
查看来源 ↗来源历史
- 收录时间
- 2026-01-20
- 索引更新时间
- 2026-01-28
- Branch
- main
收录与同步日期属于本站记录。源码修改历史请查看来源仓库。
GitHub ↗chat Comments (0)
Sign in to join the discussion and leave a comment.
No comments yet. Be the first to share your thoughts!