Byted Data Quality Review Loop
- 17 installs
- 411 repo stars
- Updated August 4, 2026
- bytedance/agentkit-samples
Byted data quality review loop is a Claude skill that enforces a strict multi-round subAgent review-and-revise loop on any deliverable before final output.
About
This skill enforces a strict multi-round review-and-revise loop on any completed deliverable before final output. It spawns an independent subAgent via sessions_spawn that returns a structured review report classifying issues as Blocker, Major, Minor or Nit. The loop repeats review and fixes until zero Blocker and zero Major issues remain, up to a default of 3 rounds. It reviews code, docs, images, audio, video and more, not just code.
- Enforces a strict multi-round subAgent review-and-revise loop before final output
- Spawns an independent reviewer via sessions_spawn that returns a structured review report with Pass/Fail and severity le
- Default gate is zero Blocker and zero Major issues, capped at 3 review rounds
Byted Data Quality Review Loop by the numbers
- 17 all-time installs (skills.sh)
- Ranked #760 of 1,352 Code Review & Quality skills by installs in the Skillselion catalog
- Data as of Aug 5, 2026 (Skillselion catalog sync)
byted-data-quality-review-loop capabilities & compatibility
No API keys required; it orchestrates a subAgent review loop.
- Capabilities
- code review · testing
- Use cases
- code review · testing
- Runs
- Runs locally
- Pricing
- Free
What byted-data-quality-review-loop says it does
For any completed deliverable, enforce a strict multi-round subAgent review-and-revise loop via sessions_spawn.
默认通过门槛：`Blocker = 0` 且 `Major = 0`
默认最多评审轮次：`3` 轮
npx skills add https://github.com/bytedance/agentkit-samples --skill byted-data-quality-review-loopAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 17 |
|---|---|
| repo stars | ★ 411 |
| Last updated | August 4, 2026 |
| Repository | bytedance/agentkit-samples ↗ |
What it does
Run a strict multi-round independent subAgent review-and-revise loop on a deliverable before final output.
Who is it for?
High-standard quality gating of any deliverable (code, docs, images, audio, video) before final output.
Skip if: Generating the content itself or defining an agent's role/persona, which stays in SOUL.md.
When should I use this skill?
You are about to deliver, submit, publish or output a final result, or the user asks for strict review or quality gating.
What you get
Drives a review->fix->re-review loop until zero Blocker and Major issues remain or the round cap is hit.
- Structured review report
- Revised deliverable that passes the gate
By the numbers
- Default max 3 review rounds
- 4 severity levels: Blocker, Major, Minor, Nit
- Pass gate: Blocker = 0 and Major = 0
Files
byted-data-quality-review-loop
将“所有已完成任务产出都必须经过严格独立复审”这一工作原则 Skill 化。其目标不是生成内容本身，而是在产出完成后，强制进入 评审 -> 修改 -> 再评审 的闭环，直到没有需要优化的问题为止。
适用场景
- 准备交付/提交/发布/输出最终结果时
- 用户明确要求“评审/挑错/过一遍/复审/质量门禁”时
- 当前执行 Agent 需要对产出做高标准、零妥协的最终把关时
输入要求
- 当前产出物（必须）：可以是代码、文档、方案、配置、脚本、图片、音频、视频、演示文稿、网页页面、数据结果、文件路径、链接等
- 任务上下文（可选但推荐）：目标、约束、受众、风格要求、已有规范
- 若缺少明确验收标准，由 subAgent 基于自身 Soul 与任务上下文自行补全本轮评审准则，并在评审报告中明确声明
核心流程（多轮闭环）
1. 当前执行 Agent 整理“当前产出版本”及可用的任务上下文。 2. 通过 sessions_spawn 启动 subAgent 进行独立评审（Round N）。 3. subAgent 输出结构化《评审报告》（见下方固定格式），明确 Pass/Fail 与问题清单；如果本轮评审准则来自其自身 Soul 与任务上下文补全，也必须在报告中显式说明。 4. 当前执行 Agent 按问题清单逐条修改，并做必要验证（按产物类型选择合适方式，例如：测试/对照示例/事实核对/一致性检查/链接可用性与引用完整性/格式与渲染检查/可复现步骤验证/音视频播放检查/字幕与画面同步检查/音频清晰度与噪声检查/封面与素材一致性检查等）。 5. 回到步骤 2，重复评审与修改，直到满足终止条件。
终止条件与保护上限
- 默认通过门槛：
Blocker = 0且Major = 0 - 默认最多评审轮次：
3轮 - 超过上限仍未通过：当前执行 Agent 必须汇报剩余问题、风险与建议方案，并请求上层流程决定：
- 继续增加轮次，或
- 接受风险并交付当前版本
subAgent 评审规范（必须遵守）
- subAgent 只做“审核把关与挑错”，不直接改产出，不替当前执行 Agent 写最终版本。
- 坚持高标准、严要求、吹毛求疵：优先发现会导致失败/回滚/误导/维护成本上升的问题。
- 不做妥协：如果存在必须修复项，结论必须为 Fail 或 Pass with fixes。
- 每条问题必须可行动：包含证据、影响、修复建议、验证方法；禁止泛泛而谈。
- subAgent 必须独立：只基于“产出物 + 上下文 + 自身 Soul 补全的评审准则”判断，不依赖当前执行 Agent 的推理过程。
- subAgent 必须按产物媒介切换评审标准：代码关注正确性与回归风险，文档关注事实、结构与可读性，图片关注信息准确性与视觉一致性，音频关注清晰度、噪声、断句、语速与口播错误，视频关注画面内容、时序、转场、字幕、音画同步与信息传达完整性。
原则边界
- 这是“工作原则 Skill”，不是角色设定 Skill，不负责定义 Agent 的职业身份或人格。
- 角色、专业身份、表达风格等内容应继续保留在各自的
SOUL.md中。 - 本 Skill 只负责注入统一的质量控制原则：凡有产出，必须经过独立挑错、多轮修正、直到达标。
严重级别说明（Severity Levels）
Blocker | 阻断级：不修复则不能交付或不能通过本轮评审的问题。Major | 重大问题：不会立刻阻断，但会显著影响质量、可信度、可用性或用户理解的问题。Minor | 次要问题：不影响主结论或主流程，但存在明显瑕疵，建议修复。Nit | 细节优化：措辞、排版、节奏、风格一致性等非关键优化项。
评审报告固定格式（subAgent 输出模板）
Review Round N
Conclusion / 结论: Pass / Pass with fixes / Fail
Severity Rule / 严重级别规则:
- Blocker = 必须修复，否则不能通过
- Major = 原则上应在通过前修复
- Minor = 建议修复
- Nit = 可选优化
Scope / 评审范围
- 本轮评审覆盖的产出类型与范围（例如：代码/文档/方案/配置/脚本/图片/音频/视频/页面；或具体文件/链接）
Must Fix / 必须修正
- [Blocker | 阻断级] 标题 / Title: ...
证据：...
影响：...
修复建议：...
验证方法：...
- [Major | 重大问题] ...
Should Fix / 建议修正
- [Minor | 次要问题] ...
Nice to Have / 可选优化
- [Nit | 细节优化] ...
Re-check Focus / 下轮复查重点
- 下一轮重点复查点（列 1-5 条）Related skills
Improve Codebase ArchitectureSafely deepen clusters of shallow modules into cohesive, testable units while respecting their external dependencies.566k196k
Caveman ReviewGet ultra-compressed, one-line code review comments that cut noise while keeping every actionable fix.270k94.7kCodebase DesignShared vocabulary for designing deep modules: improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable.266k196kCavecrewDelegate coding tasks to specialized subagents that return compressed output, keeping the main context window usable for much longer sessions.220k94.7kCode ReviewReviews a branch or PR diff on two axes at once: conformance to coding standards plus a code-smell baseline, and whether it actually implements the original spec.201k196k
Requesting Code ReviewDispatch a consistent, high-signal code reviewer subagent that catches plan deviations and quality issues before merging or continuing development.183k264kFAQ
How does the review run?
It spawns an independent subAgent via sessions_spawn that only reviews and never edits the deliverable.
What is the pass gate?
By default the deliverable passes when there are zero Blocker and zero Major issues.
How many rounds does it run?
By default up to 3 review rounds before escalating remaining issues.