Back to Details

ln-510-quality-coordinator

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.
Paths: 文件路径（
shared/
、
references/
、
../ln-*
）是相对于技能仓库根目录的。如果在当前工作目录（CWD）未找到，请定位到本SKILL.md所在目录，然后向上一级找到仓库根目录。

Quality Coordinator

质量协调器

Sequential coordinator for code quality pipeline. Invokes 4 workers in index order (511 -> 512 -> 513 -> 514) and returns aggregated results to ln-500.
代码质量流水线的顺序协调器。按索引顺序调用4个工作组件（511 -> 512 -> 513 -> 514），并将汇总结果返回至ln-500。

Purpose & Scope

目的与范围

  • Invoke ln-511-code-quality-checker (metrics, MCP Ref, static analysis)
  • Invoke ln-512-tech-debt-cleaner (auto-fix safe findings from ln-511)
  • Invoke ln-513-agent-reviewer (external agent reviews on cleaned code)
  • Run Criteria Validation (Story dependencies, AC-Task Coverage, DB Creation Principle)
  • Run linters from tech_stack.md
  • Invoke ln-514-regression-checker (test suite after all changes)
  • Return aggregated quality results to ln-500-story-quality-gate
  • No verdict determination — ln-500 decides final Gate verdict
  • 调用ln-511-code-quality-checker（指标统计、MCP参考、静态分析）
  • 调用ln-512-tech-debt-cleaner（自动修复ln-511检测出的安全可修复问题）
  • 调用ln-513-agent-reviewer（外部Agent评审清理后的代码）
  • 执行标准验证（需求依赖、AC-任务覆盖度、数据库创建原则）
  • 执行tech_stack.md中定义的代码检查工具（linters）
  • 调用ln-514-regression-checker（所有变更完成后运行测试套件）
  • 将汇总的质量结果返回至ln-500-story-quality-gate
  • 不做最终判定 —— 最终门禁判定由ln-500完成

When to Use

使用场景

  • Invoked by ln-500-story-quality-gate Phase 2
  • All implementation tasks in Story status = Done
  • 由ln-500-story-quality-gate在阶段2调用
  • 需求（Story）下的所有实现任务状态均为“完成（Done）”

Workflow

工作流程

Phase 1: Discovery

阶段1：发现

  1. Auto-discover team/config from 
    docs/tasks/kanban_board.md
  2. Load Story + task metadata from Linear (no full descriptions)
Fast-track mode: When invoked with 
--fast-track
flag (readiness 10/10), run Phase 2 with 
--skip-mcp-ref
(metrics + static only, no MCP Ref), skip Phase 3 (ln-512), Phase 4 (ln-513). Run Phase 5 (criteria), Phase 6 (linters), Phase 7 (ln-514).
Input: Story ID from ln-500-story-quality-gate
  1. 从
    docs/tasks/kanban_board.md
    自动识别团队/配置信息
  2. 从Linear加载需求（Story）及任务元数据（不包含完整描述）
快速跟踪模式： 当通过
--fast-track
参数调用（就绪度10/10）时，阶段2将使用
--skip-mcp-ref
参数运行（仅统计指标+静态分析，跳过MCP参考检查），跳过阶段3（ln-512）、阶段4（ln-513）。执行阶段5（标准验证）、阶段6（代码检查）、阶段7（ln-514）。
输入： 来自ln-500-story-quality-gate的需求ID（Story ID）

Phase 2: Code Quality (delegate to ln-511 — ALWAYS runs)

阶段2：代码质量（委托给ln-511 —— 必须执行）

MANDATORY STEP: ln-511 invocation required in ALL modes. Full gate: ln-511 runs everything (metrics + MCP Ref + static analysis). Fast-track: ln-511 runs with 
--skip-mcp-ref
(metrics + static analysis only — catches complexity, DRY, dead code without expensive MCP Ref calls).
  1. Invoke ln-511-code-quality-checker via Skill tool
    • Full: ln-511 runs code metrics, MCP Ref validation (OPT/BP/PERF), static analysis
    • Fast-track: ln-511 runs code metrics + static analysis only (skips OPT-, BP-, PERF- MCP Ref checks)
    • Returns verdict (PASS/CONCERNS/ISSUES_FOUND) + code_quality_score + issues list
  2. If ln-511 returns ISSUES_FOUND -> aggregate issues, continue (ln-500 decides action)
Invocation:
undefined
强制步骤： 无论何种模式，都必须调用ln-511。 完整门禁： ln-511执行全部检查（指标统计+MCP参考+静态分析）。 快速跟踪： ln-511执行指标统计+静态分析（跳过MCP参考中的OPT-、BP-、PERF-类检查），无需调用MCP参考接口即可捕获代码复杂度、重复代码、死代码等问题。
  1. 通过Skill工具调用ln-511-code-quality-checker
    • 完整模式：ln-511执行代码指标统计、MCP参考验证（OPT/BP/PERF）、静态分析
    • 快速跟踪模式：ln-511仅执行代码指标统计+静态分析（跳过OPT-、BP-、PERF-类MCP参考检查）
    • 返回判定结果（PASS/CONCERNS/ISSUES_FOUND）+ 代码质量得分 + 问题列表
  2. 若ln-511返回ISSUES_FOUND -> 汇总问题，继续执行（由ln-500决定后续操作）
调用方式：
undefined

Full gate:

完整门禁模式：

Skill(skill: "ln-511-code-quality-checker", args: "{storyId}")
Skill(skill: "ln-511-code-quality-checker", args: "{storyId}")

Fast-track:

快速跟踪模式：

Skill(skill: "ln-511-code-quality-checker", args: "{storyId} --skip-mcp-ref")
undefined
Skill(skill: "ln-511-code-quality-checker", args: "{storyId} --skip-mcp-ref")
undefined

Phase 3: Tech Debt Cleanup (delegate to ln-512 — SKIP if --fast-track)

阶段3：技术债务清理（委托给ln-512 —— 快速跟踪模式下跳过）

MANDATORY STEP (full gate): ln-512 invocation required. Safe auto-fixes only (confidence >=90%). Fast-track: SKIP this phase.
  1. Invoke ln-512-tech-debt-cleaner via Skill tool
    • ln-512 consumes findings from ln-511 output (passed via coordinator context)
    • Filters to auto-fixable categories (unused imports, dead code, deprecated aliases)
    • Applies safe fixes, verifies build integrity, creates commit
  2. If ln-512 returns BUILD_FAILED -> all changes reverted, aggregate issue, continue
Invocation:
Skill(skill: "ln-512-tech-debt-cleaner", args: "{storyId}")
强制步骤（完整门禁）： 必须调用ln-512。仅执行安全的自动修复（置信度≥90%）。 快速跟踪模式： 跳过此阶段。
  1. 通过Skill工具调用ln-512-tech-debt-cleaner
    • ln-512使用ln-511输出的检测结果（通过协调器上下文传递）
    • 筛选出可自动修复的类别（未使用的导入、死代码、已弃用的别名）
    • 执行安全修复，验证构建完整性，创建提交
  2. 若ln-512返回BUILD_FAILED -> 回滚所有变更，汇总问题，继续执行
调用方式：
Skill(skill: "ln-512-tech-debt-cleaner", args: "{storyId}")

Phase 4: Agent Review (delegate to ln-513 — SKIP if --fast-track)

阶段4：Agent评审（委托给ln-513 —— 快速跟踪模式下跳过）

MANDATORY STEP (full gate): ln-513 invocation required. Returns SKIPPED gracefully if agents unavailable. Fast-track: SKIP this phase.
  1. Invoke ln-513-agent-reviewer via Skill tool
    • ln-513 runs external agents (Codex + Gemini) in parallel on cleaned code
    • Critically verifies each suggestion, debates if disagreeing
    • Returns filtered suggestions with confidence scoring
  2. Merge suggestions into issues list (same prefixes: SEC-, PERF-, MNT-, ARCH-, BP-, OPT-)
  3. If verdict = SUGGESTIONS with area=security or area=correctness -> escalate aggregate to CONCERNS
Invocation:
Skill(skill: "ln-513-agent-reviewer", args: "{storyId}")
强制步骤（完整门禁）： 必须调用ln-513。若Agent不可用，将优雅返回SKIPPED状态。 快速跟踪模式： 跳过此阶段。
  1. 通过Skill工具调用ln-513-agent-reviewer
    • ln-513在清理后的代码上并行运行外部Agent（Codex + Gemini）
    • 严格验证每个建议，若存在分歧则进行校验
    • 返回带有置信度评分的筛选后建议
  2. 将建议合并到问题列表中（使用相同前缀：SEC-、PERF-、MNT-、ARCH-、BP-、OPT-）
  3. 若判定结果为SUGGESTIONS且问题领域为安全或正确性 -> 将汇总结果升级为CONCERNS
调用方式：
Skill(skill: "ln-513-agent-reviewer", args: "{storyId}")

Phase 5: Criteria Validation

阶段5：标准验证

MANDATORY READ: Load 
references/criteria_validation.md
CheckDescriptionFail Action
#1 Story DependenciesNo forward deps within Epic[DEP-] issue
#2 AC-Task CoverageSTRONG/WEAK/MISSING scoring[COV-]/[BUG-] issue
#3 DB Creation PrincipleSchema scope matches Story[DB-] issue
强制阅读： 加载
references/criteria_validation.md
检查项描述失败操作
#1 需求依赖Epic内无前置依赖标记[DEP-]问题
#2 AC-任务覆盖度评分分为STRONG/WEAK/MISSING标记[COV-]/[BUG-]问题
#3 数据库创建原则Schema范围与需求匹配标记[DB-]问题

Phase 6: Linters

阶段6：代码检查工具（Linters）

MANDATORY READ: 
shared/references/ci_tool_detection.md
(Discovery Hierarchy + Command Registry)
  1. Detect lint/typecheck commands per ci_tool_detection.md discovery hierarchy
  2. Run all detected checks (timeouts per guide: 2min linters, 5min typecheck)
  3. If any check fails -> aggregate issues, continue
强制阅读： 
shared/references/ci_tool_detection.md
（发现层级 + 命令注册表）
  1. 根据ci_tool_detection.md中的发现层级检测代码检查/类型检查命令
  2. 执行所有检测到的检查（超时时间遵循指南：代码检查2分钟，类型检查5分钟）
  3. 若任意检查失败 -> 汇总问题，继续执行

Phase 7: Regression Tests (delegate to ln-514)

阶段7：回归测试（委托给ln-514）

  1. Invoke ln-514-regression-checker via Skill tool
    • Runs full test suite, reports PASS/FAIL
    • Runs AFTER ln-512 changes to verify nothing broke
  2. If regression FAIL -> aggregate issues, continue
Invocation:
Skill(skill: "ln-514-regression-checker", args: "{storyId}")
  1. 通过Skill工具调用ln-514-regression-checker
    • 运行完整测试套件，返回PASS/FAIL结果
    • 在ln-512完成变更后执行，验证变更未引入新问题
  2. 若回归测试失败 -> 汇总问题，继续执行
调用方式：
Skill(skill: "ln-514-regression-checker", args: "{storyId}")

Phase 8: Return Results

阶段8：返回结果

Return aggregated results to ln-500:
yaml
quality_check: PASS | CONCERNS | ISSUES_FOUND
code_quality_score: {0-100}
agent_review: CODE_ACCEPTABLE | SUGGESTIONS | SKIPPED
criteria_validation: PASS | FAIL
linters: PASS | FAIL
tech_debt_cleanup: CLEANED | NOTHING_TO_CLEAN | BUILD_FAILED | SKIPPED
regression: PASS | FAIL
issues:
  - {id: "SEC-001", severity: high, finding: "...", source: "ln-511"}
  - {id: "OPT-001", severity: medium, finding: "...", source: "ln-513"}
  - {id: "DEP-001", severity: medium, finding: "...", source: "criteria"}
  - {id: "LINT-001", severity: low, finding: "...", source: "linters"}
TodoWrite format (mandatory):
- Invoke ln-511-code-quality-checker (in_progress)
- Invoke ln-512-tech-debt-cleaner (pending)
- Invoke ln-513-agent-reviewer (pending)
- Criteria Validation (Story deps, AC coverage, DB schema) (pending)
- Run linters from tech_stack.md (pending)
- Invoke ln-514-regression-checker (pending)
- Return results to ln-500 (pending)
将汇总结果返回至ln-500：
yaml
quality_check: PASS | CONCERNS | ISSUES_FOUND
code_quality_score: {0-100}
agent_review: CODE_ACCEPTABLE | SUGGESTIONS | SKIPPED
criteria_validation: PASS | FAIL
linters: PASS | FAIL
tech_debt_cleanup: CLEANED | NOTHING_TO_CLEAN | BUILD_FAILED | SKIPPED
regression: PASS | FAIL
issues:
  - {id: "SEC-001", severity: high, finding: "...", source: "ln-511"}
  - {id: "OPT-001", severity: medium, finding: "...", source: "ln-513"}
  - {id: "DEP-001", severity: medium, finding: "...", source: "criteria"}
  - {id: "LINT-001", severity: low, finding: "...", source: "linters"}
强制使用TodoWrite格式：
- Invoke ln-511-code-quality-checker (in_progress)
- Invoke ln-512-tech-debt-cleaner (pending)
- Invoke ln-513-agent-reviewer (pending)
- Criteria Validation (Story deps, AC coverage, DB schema) (pending)
- Run linters from tech_stack.md (pending)
- Invoke ln-514-regression-checker (pending)
- Return results to ln-500 (pending)

Worker Invocation (MANDATORY)

工作组件调用规则（强制）

PhaseWorkerContext
2ln-511-code-quality-checkerShared (Skill tool) — code metrics, MCP Ref, static analysis
3ln-512-tech-debt-cleanerShared (Skill tool) — auto-fix safe findings from ln-511
4ln-513-agent-reviewerShared (Skill tool) — external agent reviews on cleaned code
7ln-514-regression-checkerShared (Skill tool) — full test suite after all changes
All workers: Invoke via Skill tool — workers see coordinator context. Sequential execution: 511 -> 512 -> 513 -> 514.
Anti-Patterns:
  • Running mypy, ruff, pytest directly instead of invoking ln-511/ln-514
  • Running agent reviews directly instead of invoking ln-513
  • Auto-fixing code directly instead of invoking ln-512
  • Marking steps as completed without invoking the actual skill
  • Determining final verdict (that's ln-500's responsibility)
阶段工作组件上下文
2ln-511-code-quality-checker共享（Skill工具）—— 代码指标统计、MCP参考、静态分析
3ln-512-tech-debt-cleaner共享（Skill工具）—— 自动修复ln-511检测出的安全可修复问题
4ln-513-agent-reviewer共享（Skill工具）—— 外部Agent评审清理后的代码
7ln-514-regression-checker共享（Skill工具）—— 所有变更完成后运行完整测试套件
所有工作组件： 通过Skill工具调用 —— 工作组件可访问协调器上下文。按顺序执行：511 -> 512 -> 513 -> 514。
反模式：
  • 直接运行mypy、ruff、pytest，而非调用ln-511/ln-514
  • 直接执行Agent评审，而非调用ln-513
  • 直接自动修复代码，而非调用ln-512
  • 未实际调用对应技能即标记步骤为完成
  • 自行判定最终结果（这是ln-500的职责）

Critical Rules

关键规则

  • Return all results to ln-500; do NOT determine verdict
  • Single source of truth: rely on Linear metadata for tasks
  • Language preservation in comments (EN/RU)
  • Do not create tasks or change statuses; ln-500 decides next actions
  • 所有结果必须返回至ln-500；不得自行判定最终结果
  • 唯一可信源：依赖Linear的任务元数据
  • 注释保留原语言（EN/RU）
  • 不得创建任务或修改状态；后续操作由ln-500决定

Definition of Done

完成定义

  • ln-511 invoked (ALWAYS — full or 
    --skip-mcp-ref
    in fast-track), code quality score returned
  • ln-512 invoked (or skipped if --fast-track), tech debt cleanup results returned
  • ln-513 invoked (or skipped if --fast-track), agent review results returned
  • Criteria Validation completed (3 checks)
  • Linters executed
  • ln-514 invoked, regression results returned
  • Aggregated results returned to ln-500
  • 已调用ln-511（必须执行 —— 完整模式或快速跟踪模式下的
    --skip-mcp-ref
    ），已返回代码质量得分
  • 已调用ln-512（或在快速跟踪模式下跳过），已返回技术债务清理结果
  • 已调用ln-513（或在快速跟踪模式下跳过），已返回Agent评审结果
  • 已完成标准验证（3项检查）
  • 已执行代码检查工具
  • 已调用ln-514，已返回回归测试结果
  • 已将汇总结果返回至ln-500

Reference Files

参考文件

  • Criteria Validation: 
    references/criteria_validation.md
  • Gate levels: 
    references/gate_levels.md
  • Workers: 
    ../ln-511-code-quality-checker/SKILL.md
    , 
    ../ln-512-tech-debt-cleaner/SKILL.md
    , 
    ../ln-513-agent-reviewer/SKILL.md
    , 
    ../ln-514-regression-checker/SKILL.md
  • Caller: 
    ../ln-500-story-quality-gate/SKILL.md
  • Test planning (separate coordinator): 
    ../ln-520-test-planner/SKILL.md
  • Tech stack/linters: 
    docs/project/tech_stack.md
Version: 7.0.0 Last Updated: 2026-02-09
  • 标准验证：
    references/criteria_validation.md
  • 门禁级别：
    references/gate_levels.md
  • 工作组件：
    ../ln-511-code-quality-checker/SKILL.md
    、
    ../ln-512-tech-debt-cleaner/SKILL.md
    、
    ../ln-513-agent-reviewer/SKILL.md
    、
    ../ln-514-regression-checker/SKILL.md
  • 调用方：
    ../ln-500-story-quality-gate/SKILL.md
  • 测试规划（独立协调器）：
    ../ln-520-test-planner/SKILL.md
  • 技术栈/代码检查工具：
    docs/project/tech_stack.md
版本： 7.0.0 最后更新时间： 2026-02-09