Agent助手
- 作者仓库星标 3,626
- 作者更新于 实时读取
- 作者仓库 datadog-agent
- 领域
- AI 智能
- 兼容 Agent
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- 信任分
- 88 / 100 · 社区维护
- 作者 / 版本 / 许可
- @DataDog · 未声明 license
- Token 消耗评级
- 低消耗
- 接入复杂程度
- 即装即用
- 是否需要外部 API Key
- 不需要
- 兼容的系统
- 未声明(默认跨平台)
- 底层运行要求
- 无特殊要求
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- 网络行为
- 允许外网请求
- 安装命令数
- 26 条
档案由构建时根据 SKILL.md 与安装命令自动衍生,可能与作者实际意图存在差异。
需要注意: 未限定 allowed-tools,默认拥有全部工具权限。
---
name: agent-supply-chain-newsletter
description: Generate the Agent Supply Chain newsletter by researching team activity on GitHub and Confluence…
category: AI 智能
runtime: 无特殊运行时
---
# agent-supply-chain-newsletter 输出预览
## PART A: 任务判断
- 适用问题:提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求:目标材料、限制条件、期望输出和验收方式。
- 证据边界:围绕“Step 1: Gather team members / Step 2: Read past newsletters for format reference / Step 3: Research GitHub activity”读取原文规则,不把推断写成作者承诺。
## PART B: 执行结果
- **01** 任务判断:确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理,并标出输入、限制和预期结果。
- **02** 执行计划:优先按“Step 1: Gather team members / Step 2: Read past newsletters for format reference / Step 3: Research GitHub activity”拆成步骤,说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果:给出可复制的命令、文件改动、检查清单或内容草稿,并说明如何继续迭代。
- **04** 风险边界:结合 读取文件、写入/修改文件、会按任务需要访问外部网络、通常不需要额外 API Key 给出执行前确认项。
## Running Rules
- 读取文件、写入/修改文件;会按任务需要访问外部网络;通常不需要额外 API Key。
- 先小样例验证,再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。 原文没有稳定的斜杠命令要求。安装验证后通常全局生效,直接在对话里点名这个 Skill 并描述任务即可。
告诉 Agent 目标文件或材料、期望结果、不可改范围、是否允许联网或执行命令。本 Skill 的权限画像是:读取文件、写入/修改文件。
先用一个小任务确认它会围绕“Step 1: Gather team members / Step 2: Read past newsletters for format reference / Step 3: Research GitHub activity”工作;涉及文件或命令时,先看 diff、日志、预览或测试结果。
检查最终产物是否包含明确结果、必要证据和下一步动作;如果输出泛泛而谈,就补充输入、边界和验收标准后重跑。
---
name: agent-supply-chain-newsletter
description: Generate the Agent Supply Chain newsletter by researching team activity on GitHub and Confluence…
category: AI 智能
source: DataDog/datadog-agent
---
# agent-supply-chain-newsletter
## 什么时候使用
- 把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理,核心价值是把输入、判断、执行、验证和交付边界固定下来,避免 Agent 泛泛回答。 把任务拆成可执行、可检查…
- 面向提示词、Agent 工作流、模型评估或自动化推理,优先处理能明确输入、步骤和验收标准的工作。
## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。
## 执行规则
- 围绕「Step 1: Gather team members / Step 2: Read past newsletters for format reference / Step 3: Research GitHub activity」组织步骤,不把推断写成作者事实。
- 读取文件、写入/修改文件;会按任务需要访问外部网络;通常不需要额外 API Key。
- 先跑小样例,确认结果可检查后再扩大任务范围。
## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown,不编造命令、平台或依赖。 作者原文负责流程事实;仓库文件负责来源和命令;流狐只补充适用场景、限制和质量判断。
skill "agent-supply-chain-newsletter" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> Step 1: Gather team members / Step 2: Read past newsletters for format reference / Step 3: Research GitHub activity
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> 无特殊运行时 | 读取文件、写入/修改文件 | 会按任务需要访问外部网络
安全层 -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
输出层 -> 可复制结果 + 检查清单 + 下一步迭代
} Generate the Agent Supply Chain newsletter for the period $ARGUMENTS by researching what team members accomplished on GitHub and Confluence, then producing both a Confluence blog post draft and a Gmail draft.
Step 1: Gather team members
Fetch the current list of agent-supply-chain team members from GitHub:
gh api orgs/DataDog/teams/agent-supply-chain/members --paginate --jq '.[].login'
Ask the user if any members should be excluded (e.g. people who moved teams).
Step 2: Read past newsletters for format reference
- Search for the latest blog posts in the ASC1 Confluence space:
mcp__claude_ai_Atlassian__searchConfluenceUsingCqlwithcql:type = "blogpost" AND space = "ASC1" ORDER BY created DESC(limit 3)
- Read the most recent newsletter with
mcp__claude_ai_Atlassian__getConfluencePage(contentFormat:markdown, contentType:blog) to match its structure and tone.
The newsletter format is:
- Header: greeting, team links (ASC1, ABLD, BARX, ADX spaces), OKR link, support channel link
- "What's new?" section organized by sub-team (Agent Developer Experience, Agent Build, Agent Delivery)
- Each item has: a short title, quantified impact (time saved, percentage improvement, count), and links to PRs or docs
- "Did you know?" section with one fun/useful tip
- Footer: link to all newsletters, support channel reminder
Step 3: Research GitHub activity
For each team member, launch background agents (use run_in_background: true) to search merged PRs during the period. Split into batches of ~8 members per agent to parallelize.
Each agent should run, for every user:
# PRs in datadog-agent
gh pr list --repo DataDog/datadog-agent --author USERNAME --state merged --search "merged:START_DATE..END_DATE" --limit 50 --json title,url,mergedAt,labels
# PRs across the DataDog org (catches buildimages, k8s-ops, integrations-core, etc.)
gh search prs --author USERNAME --owner DataDog --merged --limit 20 --json repository,title,url -- "merged:START_DATE..END_DATE"
Each agent should return a summary per user, grouped thematically (build improvements, CI/CD, new features, bug fixes, etc.). Skip trivial PRs (version bumps, dependency updates). Focus on items that impact teams outside Agent Supply Chain.
Step 4: Research Confluence activity
Launch a background agent to search for relevant documentation created during the period:
mcp__claude_ai_Atlassian__searchConfluenceUsingCql
With CQL queries:
space = "ASC1" AND lastModified >= "START_DATE" AND lastModified <= "END_DATE" ORDER BY lastModified DESC(limit 25)space = "ADX" AND lastModified >= "START_DATE" AND lastModified <= "END_DATE" ORDER BY lastModified DESC(limit 25)
Identify RFCs, design docs, operational reports, and guides that are newsletter-worthy.
Step 5: Synthesize and write the newsletter
Apply the newsletter guide's filter: "Is this information impacting a team outside of the Agent Supply Chain group?" Only include items where the answer is yes.
For each item:
- Provide a quantifiable improvement (time saved, percentage change, cost reduction) when available
- Link to the relevant PR, Confluence page, or documentation
- Keep descriptions concise (2-4 sentences max per item)
Group items under:
- Agent Developer Experience (CI speed, developer tools, workflows, open source)
- Agent Build (Bazel migration, build system, platform support)
- Agent Delivery (releases, deployments, registries, security)
End with a "Did you know?" section highlighting one interesting tool, feature, or tip.
Step 6: Create outputs
6a. Confluence blog post draft
Use mcp__claude_ai_Atlassian__createConfluencePage with:
cloudId:datadoghq.atlassian.netspaceId:4662624793(ASC1 space)title:<Period> - Agent Supply Chain Monthly UpdatecontentType:blogstatus:draftcontentFormat:markdown
6b. Gmail draft
Use mcp__claude_ai_Gmail__gmail_create_draft with:
to:agent-community@datadoghq.comsubject:<Period> - Agent Supply Chain Monthly UpdatecontentType:text/html- Rich HTML body matching the Confluence content
Step 7: Present results
Return to the user:
- Links to both the Confluence draft and Gmail draft
- A bullet-point summary of the sections covered
- Remind them to:
- Review both outputs for accuracy
- Send the Gmail to themselves first to verify formatting
- Schedule the final send between 2-4pm CET, not on a Friday
- Ask for review in
#agent-devx-private/ from Damien Desmarets before publishing
先判断是否适合
作者设计意图
作者的方法与取舍
边界和复核