≈399 tokens on demand. Your agent reads this file only when SKILL.md points to it.
Investigation Budget Reference
Codebase Size Classification
| Size |
LOC |
Typical Structure |
| Small |
<5K |
Single module/app |
| Medium |
5K-50K |
Multi-module app |
| Large |
50K-500K |
Monolith or multi-service |
| XLarge |
500K+ |
Monorepo, enterprise |
Budget Allocation by Size
Small (<5K LOC)
- SCOPE: 1 pass, ≤5 min equivalent
- SURVEY: Depth 2 (entry points + direct dependencies)
- TRACE: Max 3 execution chains
- Total iterations: ≤2
Medium (5K-50K LOC)
- SCOPE: 1-2 passes, boundary identification first
- SURVEY: Depth 3 (entry + deps + cross-module)
- TRACE: Max 5 execution chains
- Total iterations: ≤3
Large (50K-500K LOC)
- SCOPE: 2+ passes, boundary-first approach mandatory
- SURVEY: Depth 4+ (progressive disclosure)
- TRACE: Max 8 execution chains, prioritize critical path
- Total iterations: ≤4
- Trigger: Ripple co-use recommended when affected services ≥3
XLarge (500K+ LOC)
- SCOPE: Component-level decomposition required
- SURVEY: Per-component deep dive, cross-component shallow
- TRACE: Max 10 chains, strict path prioritization
- Total iterations: ≤5
- Trigger: Sherpa decomposition mandatory
Context Budget Management
- Track token consumption per phase
- Abandon TRACE if >60% of context budget consumed in SCOPE+SURVEY
- Summarize findings before deep-diving (avoid context eviction of early findings)
Escalation on Budget Exhaustion
- Summarize current findings
- Identify remaining unknowns
- Handoff to Scout (if bug suspected) or Sherpa (if decomposition needed)
