AgentSkillsCN

breadth-of-thought

全面探索解决方案空间。适用于需要穷举所有可行选项（而不仅仅是最佳方案）、解决方案空间尚不明确，或无法承受错过任何替代方案时使用。

SKILL.md
--- frontmatter
name: breadth-of-thought
description: Exhaustive solution space exploration. Use when you need ALL viable options (not just the best), solution space is unknown, or you can't afford to miss alternatives.
metadata: {"clawdbot":{"requires":{"env":[]}}}
user-invocable: true

Breadth of Thought (BoT)

Exhaustive exploration methodology for discovering ALL viable solutions.

When to Use

  • •Solution space is unknown or poorly defined
  • •Need multiple viable options (not just one best)
  • •Can't afford to miss alternatives
  • •Exploring possibilities before committing
  • •Brainstorming or ideation phases

Key Difference from ToT

ToTBoT
Find THE bestFind ALL viable
Prune aggressively (keep top 2-3)Prune conservatively (keep >40% confidence)
Deep exploration of winnersBroad exploration of everything
Single recommendationMultiple options with trade-offs

Process

Phase 1: Divergent Generation

Generate 8-10 approaches at each level:

  • •Don't filter yet
  • •Include unconventional ideas
  • •Cross-domain analogies welcome
  • •Quantity over quality initially

Phase 2: Conservative Filtering

Keep anything above 40% viability threshold:

  • •"Might work" = keep
  • •"Definitely won't work" = prune
  • •Document why each was kept/pruned

Phase 3: Parallel Expansion

For EACH remaining option, generate sub-approaches:

code
Option 1 → 1a, 1b, 1c
Option 2 → 2a, 2b, 2c
Option 3 → 3a, 3b, 3c
...

Phase 4: Viability Assessment

Score all remaining options:

  • •Feasibility (1-10)
  • •Uniqueness (how different from others)
  • •Trade-off profile (what you gain/lose)

Phase 5: Portfolio Output

Return 3-5 viable solutions with:

  • •Clear differentiation
  • •Trade-off comparison
  • •Use case fit

Output Template

markdown
## BoT Exploration: [Problem]

### Solution Portfolio

#### Option 1: [Name]
- **Approach**: [description]
- **Best for**: [use case]
- **Trade-offs**: [gains vs losses]
- **Confidence**: X%

#### Option 2: [Name]
- **Approach**: [description]
- **Best for**: [use case]
- **Trade-offs**: [gains vs losses]
- **Confidence**: X%

#### Option 3: [Name]
...

### Comparison Matrix
| Option | Speed | Cost | Complexity | Risk | Best When |
|--------|-------|------|------------|------|-----------|
| 1 | | | | | |
| 2 | | | | | |
| 3 | | | | | |

### Eliminated Options
- [Option X]: [why not viable]

### Recommendation
No single "best" - choose based on priorities:
- If [priority A] matters most → Option 1
- If [priority B] matters most → Option 2
- If [priority C] matters most → Option 3

Example

Problem: Data pipeline architecture options

BoT Output:

  1. •Batch ETL - Best for: cost-sensitive, latency-tolerant
  2. •Stream Processing - Best for: real-time requirements
  3. •Lambda Architecture - Best for: mixed workloads
  4. •Lakehouse - Best for: unified analytics + ML
  5. •Event Sourcing - Best for: audit trails, replay needs

Each is viable - choice depends on specific requirements.