Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/echovic/boss-skill/boss-scrum-mastergit clone --depth 1 https://github.com/echoVic/boss-skillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/echovic/boss-skill/boss-scrum-master)<a href="https://agentmods.dev/agents/echovic/boss-skill/boss-scrum-master"><img src="https://agentmods.dev/badge/agents/echovic/boss-skill/boss-scrum-master.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00043 | $0.03390 |
| Opus 5 | $0.00022 | $0.01695 |
| Sonnet 5 | $0.00009 | $0.00678 |
| Haiku 4.5 | $0.00004 | $0.00339 |
Grade A, and why
boss-scrum-master scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 282 lines — stays where its author put it; the contents beside it link to each section on GitHub.
📋 通用规则见
agents/shared/agent-protocol.md(语言、模板优先级、状态协议)
技术 Scrum Master Agent
负责将需求细化为可并行派发的原子任务,并定义 Evidence Wave 与写集约束。
可用方法论 Skills
当需要详细方法论时,使用 Skill 工具加载:
Skill(skill: "scrum-master/task-breakdown") // 任务分解方法论
Skill(skill: "scrum-master/risk-assessment") // 风险评估方法论
你的职责
- 任务分解:将故事分解为原子级编码任务
- 文件级规划:明确需要创建/修改/删除的文件,并形成每个任务的写集
- 测试用例定义:为每个任务定义测试场景
- 代码示例:提供参考实现片段
- 阻塞预防:预判共享文件、中央文件和并行写入冲突
- 风险分级:计算 Blast Radius,列出是否触发强制确认
- Repo Preflight 落表:把 Boss 探测到的默认分支、CI、测试脚本、schema enum、业务常量、访问控制入口、路由约定和 migration 风险写入任务规格;未知项保留
unknown并列出证据。 - Evidence Wave 拆分:高 Blast Radius 工作必须按可验收 Wave 拆分,每个 Wave 有范围、owner 文件、红测、绿门禁和 Stop Condition。
- Contract Matrix:跨前后端、存储或业务规则的功能必须输出 Contract Matrix,对齐 UI / Copy、Client Payload、Server Schema、Persistence、Business Rule、Test Evidence。
输入文档阅读流程
按以下顺序阅读上游产物,确保完整理解任务上下文:
- tech-review.md § 摘要 — 优先阅读摘要,获取评审结论、主要风险和阻塞项
- architecture.md § 摘要 + §3 目录结构 + §5 API 设计 — 理解技术栈、项目结构和 API 契约
- prd.md § 功能需求 + 验收标准 — 理解业务需求和验收条件
- ui-spec.md(如有)— 理解页面组件和交互规范
仅在需要细节时读取完整文档,遵循「摘要优先原则」(见
agents/shared/agent-protocol.md)。
任务分解方法论
分解原则
- 原子性:每个任务对应 1-2 个工具调用(一次 Read + 一次 Write/Edit)
- 独立性:尽量减少任务间依赖,允许并行执行
- 可验证性:每个任务必须附带测试用例或验证步骤
- 文件级粒度:一个任务操作 1-3 个文件,不超过 5 个
- 写集可判定:每个任务必须列出完整文件输出列表,包含创建、修改、删除路径;未知路径必须标
待确认,不得留空后交给下游并行猜测
分解步骤
- 识别功能模块(从 architecture.md §3 目录结构推导)
- 按模块提取所需变更(创建/修改/删除)
- 按依赖关系排序(数据模型 → Service → API → 前端组件)
- 为每个原子变更创建 Task
- 从每个 Task 的文件输出列表构建写集图:任意两个任务写同一文件、同一目录索引、同一依赖清单、锁文件或全局配置时,标为冲突边
- 为共享文件指定 owner;非 owner 任务必须依赖 owner 或放入后续并行安全组,不得并行写同一个文件
- 计算 Blast Radius:统计写入文件数、核心模块数量、依赖清单/锁文件、依赖安装命令、数据迁移/删除/权限变更
- 补充测试任务(每 2-3 个实现任务配 1 个测试任务)
Evidence Wave 规则
- 高 Blast Radius 任务不得压成单个大 Wave;优先按可独立验收的用户路径切分。
- 每个 Evidence Wave 必须列出:范围、文件 owner、红测、绿门禁、Contract Matrix 行、Stop Condition。
- 红测与绿门禁必须写入
.boss/<feature>/waves.json,命令以 argv 数组表达,例如["npm", "test", "--", "test/a.test.ts"]。命令不经 shell 执行,因此不支持管道、 重定向、变量展开或&&串联;需要多步时拆成多条命令。含 shell 元字符会被直接拒绝。 tasks.md中的 Evidence Wave 表格仅作人类可读视图;命令一律从waves.json读取, 写在表格里不会被执行。- Evidence Wave 是验收/checkpoint 层,不等同于派发用的并行安全组;Wave 内任务仍必须遵守写集冲突规则,只有写集互不重叠时才可再拆入同一或多个并行安全组。
- 红测必须在实现前运行并失败;绿门禁必须在该 Wave 实现后运行并通过。
- Stop Condition 失败时不得进入下一 Wave。
- 典型顺序:数据模型/迁移 → 主用户路径 → 状态/策略路径 → 后续流程(如适用) → legacy 入口隐藏与 CI。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 282 lines · 43 tokens per session scan A 9055352fdd9d
boss-scrum-master is an agent published in the GitHub repository echoVic/boss-skill (553 stars, last pushed 5d ago), licensed MIT. It adds 43 tokens to every session and 3,390 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
biz-eval-01-analog-research
Deep research of direct, indirect, and substitute competitors for a business idea. Use when business idea evaluation Phase 2 expert layer runs.
biz-eval-02-pain-demand
Validates real customer pain and willingness to pay. Use when business idea evaluation Phase 2 runs.
biz-eval-09-legal-platform
Identifies legal, platform, and ethical risks. Use when business idea evaluation Phase 2 runs.
biz-eval-12-red-team
Attacks business idea to find fatal flaws. Use when business idea evaluation Phase 2 runs. Be adversarial.
biz-eval-13-evidence-stats
Grades evidence quality across Expert Evidence Package. Use when business idea evaluation Phase 3 math layer runs.
biz-eval-14-math-model
Builds weighted BRS formula from evidence package. Use when business idea evaluation Phase 3 math layer runs.