ln-811-algorithm-optimizer

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Paths: File paths (
shared/
,
references/
,
../ln-*
) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root.

路径说明： 文件路径（
shared/
、
references/
、
../ln-*
）均相对于技能仓库根目录。如果在当前工作目录（CWD）未找到，请定位到本SKILL.md所在目录，再向上一级即为仓库根目录。

ln-811-algorithm-optimizer

Type: L3 Worker Category: 8XX Optimization Parent: ln-810-performance-optimization-coordinator

Optimizes target function performance via autoresearch loop: establish baseline benchmark, research best practices, generate 3-7 hypotheses, apply each with keep/discard verification.

Type: L3 Worker Category: 8XX Optimization Parent: ln-810-performance-optimization-coordinator

通过自动研究循环优化目标函数性能：建立基准测试基线、研究最佳实践、生成3-7个优化假设、逐一验证并决定保留/丢弃方案。

Overview

概述

Aspect	Details
Input	`target_file` + `target_function` (or audit findings from ln-650)
Output	Optimized function with benchmark proof, optimization report
Pattern	Autoresearch: modify → benchmark → keep (≥10%) / discard

维度	详情
输入	`target_file` + `target_function` （或来自ln-650的审计结果）
输出	附带基准测试证明的优化后函数、优化报告
模式	自动研究循环：修改→基准测试→保留（提升≥10%）/丢弃

Workflow

工作流

Phases: Pre-flight → Baseline → Research → Hypothesize → Optimize Loop → Report

阶段： 前期检查→基准建立→研究分析→假设生成→优化循环→结果报告

Phase 0: Pre-flight Checks

阶段0：前期检查

Check	Required	Action if Missing
Target file exists	Yes	Block optimization
Target function identifiable	Yes	Block optimization
Test infrastructure	Yes	Block optimization (see ci_tool_detection.md)
Test coverage for target function	Yes	Block — no coverage = no safety net
Git clean state	Yes	Block (need clean baseline for revert)
Benchmark infrastructure	No	Generate benchmark (see references)

MANDATORY READ: Load

shared/references/ci_tool_detection.md

— use Benchmarks + Test Frameworks sections.

检查项	是否必填	缺失时操作
目标文件存在	是	阻止优化
目标函数可识别	是	阻止优化
测试基础设施就绪	是	阻止优化（参见ci_tool_detection.md）
目标函数有测试覆盖率	是	阻止优化——无覆盖率=无安全保障
Git处于干净状态	是	阻止优化（需要干净基线用于回滚）
基准测试基础设施就绪	否	生成基准测试（参见参考文档）

必读内容： 加载

shared/references/ci_tool_detection.md

文档——使用其中的基准测试和测试框架章节内容。

Coverage Verification

测试覆盖率验证

Before starting optimization, verify target function has test coverage:

Step	Action
1	Grep test files for target function name / imports from target module
2	If ≥1 test references target → PROCEED
3	If 0 tests reference target → BLOCK with "no test coverage for {function}"

Without test coverage, benchmark improvements are meaningless — the optimized function may produce wrong results faster.

开始优化前，需验证目标函数是否有测试覆盖率：

步骤	操作
1	在测试文件中搜索目标函数名称/来自目标模块的导入语句
2	如果≥1个测试用例引用目标函数→继续执行
3	如果0个测试用例引用目标函数→阻止优化，提示“{function}无测试覆盖率”

没有测试覆盖率的话，基准测试的性能提升毫无意义——优化后的函数可能只是更快地输出错误结果。

Worktree & Branch Isolation

工作树与分支隔离

MANDATORY READ: Load

shared/references/git_worktree_fallback.md

— use ln-811 row.

All work (edits, benchmarks, KEEP commits) in worktree. Never modify main worktree.

必读内容： 加载

shared/references/git_worktree_fallback.md

文档——使用其中ln-811对应的行内容。

所有工作（编辑、基准测试、保留操作的提交）都在独立工作树中进行。严禁修改主工作树。

Phase 1: Establish Baseline

阶段1：建立基准基线

Step 1.1: Detect or Generate Benchmark

步骤1.1：检测或生成基准测试

Situation	Action
Existing benchmark found	Use as-is
No benchmark exists	Generate minimal benchmark (see benchmark_generation.md)

场景	操作
找到现有基准测试	直接使用
无现有基准测试	生成最小化基准测试（参见benchmark_generation.md）

Step 1.2: Run Baseline

步骤1.2：运行基准测试

Parameter	Value
Runs	5
Metric	Median execution time
Warm-up	1 discarded run
Output	`baseline_median` , `baseline_p95`

Save baseline result — all improvements measured against this.

参数	取值
运行次数	5
衡量指标	执行时间中位数
预热次数	丢弃1次预热运行结果
输出结果	`baseline_median` , `baseline_p95`

保存基准测试结果——所有性能提升都将以此为参照。

Phase 2: Research Best Practices

阶段2：研究最佳实践

MANDATORY READ: Load

shared/references/research_tool_fallback.md

for MCP tool chain.

必读内容： 加载

shared/references/research_tool_fallback.md

文档以获取MCP工具链信息。

Research Strategy

研究策略

Priority	Tool	Query Template
1	mcp__context7__query-docs	`"{language} {algorithm_type} optimization techniques"`
2	mcp__Ref__ref_search_documentation	`"{language} {function_name} performance best practices"`
3	WebSearch	`"{algorithm_type} optimization {language} benchmark {current_year}"`

优先级	工具	查询模板
1	mcp__context7__query-docs	`"{language} {algorithm_type} optimization techniques"`
2	mcp__Ref__ref_search_documentation	`"{language} {function_name} performance best practices"`
3	WebSearch	`"{algorithm_type} optimization {language} benchmark {current_year}"`

Research Output

研究输出

Collect optimization techniques applicable to the target function. For each technique note:

Name and description
Expected improvement category (algorithmic complexity, memory, cache, parallelism)
Applicability conditions (data size, structure, language features)

收集适用于目标函数的优化技术。对于每种技术，需记录：

技术名称与描述
预期提升类别（算法复杂度、内存、缓存、并行化）
适用条件（数据规模、结构、语言特性）

Phase 3: Generate Hypotheses (3-7)

阶段3：生成优化假设（3-7个）

Hypothesis Sources

假设来源

MANDATORY READ: Load optimization_categories.md for category checklist.

Source	Priority
Research findings (Phase 2)	1
Optimization categories checklist	2
Code analysis (anti-patterns in target)	3

必读内容： 加载optimization_categories.md文档获取优化类别清单。

来源	优先级
阶段2的研究成果	1
优化类别清单	2
代码分析（目标函数中的反模式）	3

Hypothesis Format

假设格式

Field	Description
id	H1, H2, ... H7
category	From optimization_categories.md
description	What to change
expected_impact	Estimated improvement %
risk	Low / Medium / High
dependencies	Other hypotheses this depends on

字段	描述
id	H1, H2, ... H7
category	来自optimization_categories.md
description	需修改的内容
expected_impact	预估提升百分比
risk	Low / Medium / High
dependencies	依赖的其他假设

Ordering

排序规则

Sort by:

expected_impact DESC, risk ASC

. Independent hypotheses first (no dependencies).

排序依据：

预期影响降序，风险升序

。优先处理独立假设（无依赖项）。

Phase 4: Optimize Loop (Keep/Discard)

阶段4：优化循环（保留/丢弃）

Per-Hypothesis Cycle

单假设循环流程

FOR each hypothesis (H1..H7):
  1. APPLY: Edit target function (surgical change, function body only)
  2. VERIFY: Run tests
     IF tests FAIL (assertion) → DISCARD (revert) → next hypothesis
     IF tests CRASH (runtime error, OOM, import error):
       IF fixable (typo, missing import) → fix & re-run ONCE
       IF fundamental (design flaw, incompatible API) → DISCARD + log "crash: {reason}"
  3. BENCHMARK: Run 5 times, take median
  4. COMPARE: improvement = (baseline - new) / baseline * 100
     IF improvement >= 10% → KEEP:
       git add {target_file}
       git commit -m "perf(H{N}): {description} (+{improvement}%)"
       new baseline = new median
     IF improvement < 10%  → DISCARD (revert edit)
  5. LOG: Record result to experiment log + report

FOR each hypothesis (H1..H7):
  1. APPLY: Edit target function (surgical change, function body only)
  2. VERIFY: Run tests
     IF tests FAIL (assertion) → DISCARD (revert) → next hypothesis
     IF tests CRASH (runtime error, OOM, import error):
       IF fixable (typo, missing import) → fix & re-run ONCE
       IF fundamental (design flaw, incompatible API) → DISCARD + log "crash: {reason}"
  3. BENCHMARK: Run 5 times, take median
  4. COMPARE: improvement = (baseline - new) / baseline * 100
     IF improvement >= 10% → KEEP:
       git add {target_file}
       git commit -m "perf(H{N}): {description} (+{improvement}%)"
       new baseline = new median
     IF improvement < 10%  → DISCARD (revert edit)
  5. LOG: Record result to experiment log + report

Safety Rules

安全规则

Rule	Description
Scope	Only target function body; no signature changes
Dependencies	No new package installations
Revert	`git checkout -- {target_file}` on discard
Time budget	30 minutes total for all hypotheses
Compound	Each KEEP becomes new baseline for next hypothesis
Traceability	Each KEEP = separate git commit with hypothesis ID in message
Isolation	All work in isolated worktree; never modify main worktree

规则	描述
Scope	仅修改目标函数体；不修改函数签名
Dependencies	禁止安装新依赖包
Revert	丢弃时执行 `git checkout -- {target_file}`
Time budget	所有假设的总时间预算为30分钟
Compound	每次保留的优化将作为下一个假设的新基准
Traceability	每次保留操作对应独立的Git提交，提交信息中包含假设ID
Isolation	所有工作在独立工作树中进行；严禁修改主工作树

Keep/Discard Decision

保留/丢弃决策

Condition	Decision	Action
Tests fail	DISCARD	Revert, log reason
Improvement ≥ 10%	KEEP	Update baseline
Improvement 10-20% BUT complexity increase	REVIEW	Log as "marginal + complex", prefer DISCARD
Improvement < 10%	DISCARD	Revert, log as "insufficient gain"
Regression (slower)	DISCARD	Revert, log regression amount

Simplicity criterion (per autoresearch): If improvement is marginal (10-20%) and change significantly increases code complexity (>50% more lines, deeply nested logic, hard-to-read constructs), prefer DISCARD. Simpler code at near-equal performance wins.

条件	决策	操作
测试失败	DISCARD	回滚，记录原因
性能提升≥10%	KEEP	更新基准基线
性能提升10-20%但代码复杂度增加	REVIEW	记录为“提升有限+复杂度高”，优先选择丢弃
性能提升<10%	DISCARD	回滚，记录为“提升幅度不足”
性能退化（运行更慢）	DISCARD	回滚，记录退化幅度

简洁性准则（自动研究循环要求）： 如果性能提升幅度有限（10-20%）且修改导致代码复杂度显著增加（代码行数增加>50%、深层嵌套逻辑、难以阅读的结构），则优先选择丢弃。在性能接近的情况下，更简洁的代码更优。

Phase 5: Report Results

阶段5：报告结果

Report Schema

报告模板

Field	Description
target	File path + function name
baseline	Original median benchmark
final	Final median after all kept optimizations
total_improvement	Percentage improvement
hypotheses_tested	Total count
hypotheses_kept	Count of kept optimizations
hypotheses_discarded	Count + reasons
optimizations[]	Per-kept: id, category, description, improvement%

字段	描述
target	文件路径+函数名称
baseline	原始基准测试中位数
final	所有保留优化后的最终中位数
total_improvement	总提升百分比
hypotheses_tested	测试的假设总数
hypotheses_kept	保留的优化假设数量
hypotheses_discarded	丢弃的数量+原因
optimizations[]	每个保留项的：id、类别、描述、提升百分比

Experiment Log

实验日志

Write to

{project_root}/.optimization/ln-811-log.tsv

Column	Description
timestamp	ISO 8601
hypothesis_id	H1..H7
category	From optimization_categories.md
description	What changed
baseline_ms	Baseline median before this hypothesis
result_ms	New median after change
improvement_pct	Percentage change
status	keep / discard / crash
commit	Git commit hash (if kept)

Append to existing file if present (enables tracking across multiple runs).

写入至

{project_root}/.optimization/ln-811-log.tsv

：

列名	描述
timestamp	ISO 8601格式时间戳
hypothesis_id	H1..H7
category	来自optimization_categories.md
description	修改内容
baseline_ms	本次假设前的基准中位数（毫秒）
result_ms	修改后的新中位数（毫秒）
improvement_pct	提升百分比
status	keep / discard / crash
commit	Git提交哈希（若保留）

如果文件已存在则追加内容（支持跨多次运行的跟踪）。

Cleanup

清理操作

Action	When
Remove generated benchmark	If benchmark was auto-generated AND no kept optimizations
Keep generated benchmark	If any optimization was kept (proof of improvement)

操作	执行时机
删除自动生成的基准测试	当基准测试为自动生成且无保留的优化方案时
保留自动生成的基准测试	当有任何优化方案被保留时（作为性能提升的证明）

Configuration

配置项

yaml

Options:
  # Target
  target_file: ""
  target_function: ""

  # Benchmark
  benchmark_runs: 5
  improvement_threshold: 10    # percent
  warmup_runs: 1

  # Hypotheses
  max_hypotheses: 7
  min_hypotheses: 3

  # Safety
  time_budget_minutes: 30
  allow_new_deps: false
  scope: "function_body"       # function_body | module

  # Verification
  run_tests: true
  run_lint: false

yaml

Options:
  # Target
  target_file: ""
  target_function: ""

  # Benchmark
  benchmark_runs: 5
  improvement_threshold: 10    # percent
  warmup_runs: 1

  # Hypotheses
  max_hypotheses: 7
  min_hypotheses: 3

  # Safety
  time_budget_minutes: 30
  allow_new_deps: false
  scope: "function_body"       # function_body | module

  # Verification
  run_tests: true
  run_lint: false

Error Handling

错误处理

Error	Cause	Solution
No benchmark framework	Stack not in ci_tool_detection.md	Generate inline benchmark
All hypotheses discarded	No effective optimization found	Report "no improvements found"
Benchmark noise too high	Inconsistent timing	Increase runs to 10, use p50
Test flake	Non-deterministic test	Re-run once; if flakes again, skip hypothesis

错误类型	原因	解决方案
无基准测试框架	技术栈未在ci_tool_detection.md中列出	生成内置基准测试
所有假设均被丢弃	未找到有效优化方案	报告“未找到可提升的优化点”
基准测试结果波动过大	计时结果不一致	将运行次数增加至10次，使用p50值
测试用例不稳定	测试结果非确定性	重新运行一次；若仍不稳定，跳过该假设

References

参考文档

benchmark_generation.md
optimization_categories.md
```
shared/references/ci_tool_detection.md
```
(Benchmarks section)

shared/references/research_tool_fallback.md

benchmark_generation.md
optimization_categories.md
```
shared/references/ci_tool_detection.md
```
（基准测试章节）

shared/references/research_tool_fallback.md

Definition of Done

完成标准

Test coverage for target function verified before optimization
Target function identified and baseline benchmark established (5 runs, median)
Research completed via MCP tool chain (Context7/Ref/WebSearch)
3-7 hypotheses generated, ordered by expected impact
Each hypothesis tested: apply → tests → benchmark → keep/discard
Each kept optimization = separate git commit with hypothesis ID
Kept optimizations compound (each becomes new baseline)
Marginal gains (10-20%) with complexity increase reviewed via simplicity criterion
Tests pass after all kept optimizations
Experiment log written to
```
.optimization/ln-811-log.tsv
```
Report returned with baseline, final, improvement%, per-hypothesis results
Generated benchmark cleaned up if no optimizations kept
All changes on isolated branch, pushed to remote

Version: 1.0.0 Last Updated: 2026-03-08

优化前已验证目标函数的测试覆盖率
已识别目标函数并建立基准测试基线（5次运行，取中位数）
通过MCP工具链完成研究分析
已生成3-7个优化假设，按预期影响和风险排序
每个假设均已完成测试：修改→测试→基准测试→保留/丢弃
每次保留的优化对应独立的Git提交，提交信息包含假设ID
保留的优化将作为后续假设的新基准
对提升幅度有限（10-20%）且增加代码复杂度的优化已按简洁性准则评估
所有保留优化完成后测试通过
实验日志已写入
```
.optimization/ln-811-log.tsv
```
已返回包含基线、最终结果、总提升、各假设结果的报告
若无优化被保留，已清理自动生成的基准测试
所有修改均在独立分支完成并推送至远程仓库

Version: 1.0.0 Last Updated: 2026-03-08