bobo-ai-cli 1.0.1

package/knowledge/rules/performance.md

@@ -0,0 +1,101 @@
+---
+id: "performance"
+title: "Performance Optimization"
+category: "root"
+tags: ["performance optimization", "model selection strategy", "context window management", "ultrathink + plan mode", "build troubleshooting"]
+source: "C:/Users/Administrator/.claude/rules/performance.md"
+---
+
+# Performance Optimization
+
+> **新增**: Cache Management (2026-04-01) — 详见 `domain/cache-management.md`
+
+## Cache & Token Optimization (优先级最高)
+
+**立即应用**:
+
+1. **使用 cc-cache-fix** — 安装 `claude-patched` 替代 `claude`
+   - 补丁1: Delta 附件持久化 → +30% 缓存命中
+   - 补丁2: Hash 稳定性 → +20% 缓存命中
+   - 补丁3: TTL 1小时 → +15% 缓存命中
+   - **总计**: ~60% token 节省
+
+2. **按需加载 Context** — 不全量加载 CLAUDE.md
+   - Layer 0: 核心规则 (20KB)
+   - Layer 1: Task Router (3KB)
+   - Layer 2: 相关能力 (15-30KB)
+   - Layer 3: 具体案例 (按需)
+
+3. **会话分割** — 长任务 (>1小时) 分成 3-5 个短会话
+   - 每个会话独立缓存
+   - 使用 `task_plan.md` 保存进度
+   - 总 token 节省 30-40%
+
+4. **缓存预热** — 会话开始时加载常用上下文
+   - 加载 CLAUDE.md + 核心规则 + 记忆
+   - 预计算常见操作
+   - 首次操作 token -30%
+
+**验证效果**:
+```bash
+claude-patched
+python cc-cache-fix/test_cache.py
+python cc-cache-fix/usage_audit.py
+```
+
+**预期结果**: 60-70% token 节省（相对基准）
+
+---
+
+## Model Selection Strategy
+
+**Haiku 4.5** (90% of Sonnet capability, 3x cost savings):
+
+- Lightweight agents with frequent invocation
+- Pair programming and code generation
+- Worker agents in multi-agent systems
+
+**Sonnet 4.5** (Best coding model):
+
+- Main development work
+- Orchestrating multi-agent workflows
+- Complex coding tasks
+
+**Opus 4.5** (Deepest reasoning):
+
+- Complex architectural decisions
+- Maximum reasoning requirements
+- Research and analysis tasks
+
+## Context Window Management
+
+Avoid last 20% of context window for:
+
+- Large-scale refactoring
+- Feature implementation spanning multiple files
+- Debugging complex interactions
+
+Lower context sensitivity tasks:
+
+- Single-file edits
+- Independent utility creation
+- Documentation updates
+- Simple bug fixes
+
+## Ultrathink + Plan Mode
+
+For complex tasks requiring deep reasoning:
+
+1. Use `ultrathink` for enhanced thinking
+2. Enable **Plan Mode** for structured approach
+3. "Rev the engine" with multiple critique rounds
+4. Use split role sub-agents for diverse analysis
+
+## Build Troubleshooting
+
+If build fails:
+
+1. Use **build-error-resolver** agent
+2. Analyze error messages
+3. Fix incrementally
+4. Verify after each fix