Skip to content
Back to search

claude-em

SkillWorks with Claude CodeGitHub

Turns Claude Code into an Engineering Manager operating system: your team's context plus skills for planning, epics, 1:1s and performance reviews.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked yesterday
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/jcesarperez--claude-em/badge.svg)](/listings/jcesarperez--claude-em)

HTML

<a href="/listings/jcesarperez--claude-em"><img src="/api/listings/jcesarperez--claude-em/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

1. Click on **[Use this template](https://github.com/new?template_name=claude-em&template_owner=jcesarperez)** and create your own repository

2. Install Claude Code:

npm install -g @anthropic-ai/claude-code

3. Add your EM style and team context:

* Copy `data/em_style_example.md` → `data/em_style.md` and fill in your style * Copy `data/team_example.md` → `data/team_myteam.md` and fill in your team and members

4. Open Claude in your workspace:

claude

5. Try your first prompt:

> I want to write a discovery epic to assess frontend performance issues

6. Complete the full setup to unlock data-driven workflows (Jira, GitHub, and more)

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 354 lines
  1. # Claude EM
  2. **Claude EM turns Claude into an Engineering Manager operating system.**
  3. Not a generic AI. Not a blank prompt.
  4. A context-aware, opinionated system that understands:
  5. - your team
  6. - your context
  7. - how you run engineering
  8. ## Why Claude EM
  9. It comes with:
  10. 1. Persistent context → **your team** + **your way of working**
  11. 2. Reusable workflows → **skills** that support real EM tasks
  12. Claude EM is built for **day-to-day EM work**:
  13. - Plan initiatives with clear scope and trade-offs
  14. - Write epics and stories that engineers can actually build
  15. - Analyze IC performance using real data (Jira + GitHub)
  16. - Prepare 1:1s with context and intent
  17. - Think through org and leadership decisions
  18. It is intentionally **CLI-first**: faster, more predictable, and more cost-efficient (tokens).
  19. Works out of the box with **Jira** and **GitHub**. Other tools can be added via CLI scripts or MCP servers.
  20. ---
  21. ## Quick start (2 minutes)
  22. 1. Click on **[Use this template](https://github.com/new?template_name=claude-em&template_owner=jcesarperez)** and create your own repository
  23. 2. Install Claude Code:
  24. ```bash
  25. npm install -g @anthropic-ai/claude-code
  26. ```
  27. 3. Add your EM style and team context:
  28. * Copy `data/em_style_example.md` → `data/em_style.md` and fill in your style
  29. * Copy `data/team_example.md` → `data/team_myteam.md` and fill in your team and members
  30. 4. Open Claude in your workspace:
  31. ```bash
  32. claude
  33. ```
  34. 5. Try your first prompt:
  35. > I want to write a discovery epic to assess frontend performance issues
  36. 6. Complete the full setup to unlock data-driven workflows (Jira, GitHub, and more)
  37. ---
  38. ## Example use cases
  39. * Plan an initiative
  40. > We have recurring CI flakiness slowing down delivery. Plan an initiative to reduce it
  41. * Analyze IC activity
  42. > Analyze Maria's last 30 days
  43. * Prepare a 1:1
  44. > Prepare a 1:1 with Ana. She’s delivering less lately and seems disengaged. I want to understand what's going on
  45. * Write a strategy
  46. > Write a strategy to reduce operational load on the team while maintaining reliability
  47. * Mentor
  48. > I'm considering reorganizing the team into two squads. Help me think through trade-offs and risks
  49. * Prepare the daily meeting
  50. > Give me a daily-ready view of the team's board
  51. * Generate a user story map
  52. > Generate a user story map from this PRD: [paste PRD]
  53. * Check an epic
  54. > Check this epic
  55. * Write a delivery epic
  56. > Write an epic to implement rate limiting in our API. Include scope, risks, and definition of done
  57. * Write a discovery epic
  58. > We have performance issues in the frontend. Write a discovery epic to understand root causes and options
  59. * Write a weekly update
  60. > Write the weekly update for the platform team
  61. > Write the project update for the migration initiative
  62. ---
  63. ## How it works
  64. * **CLAUDE.md** defines how Claude should behave
  65. * **EM style** (`data/em_style.md`) defines your personal way of working
  66. * **Skills** (`.claude/skills/`) are reusable prompts for real EM tasks
  67. * **Team context** (`data/team_*.md`) defines your team
  68. * **Data** (`data/`) stores shared information (Jira, GitHub, etc.)
  69. * **Initiatives** (`initiatives/`) are folders where work, analysis, and outputs live
  70. * **CLI-first approach**: prefers CLI tools over MCP servers to stay fast, predictable, and low-cost in terms of tokens
  71. ---
  72. ## Skills
  73. | Skill | What it does |
  74. |---|---|
  75. | `/check-board` | Daily health check of a Jira Kanban board: flow, people load, stuck/blocked issues, and bugs at risk of breaching SLA |
  76. | `/check-epic` | Analyze an Epic's definition, decomposition, and execution using Jira data |
  77. | `/check-ic-activity` | Analyze an IC's delivery, quality, and collaboration using Jira + GitHub data |
  78. | `/jira` | Create and query Jira issues — epics, stories, bugs, and tasks |
  79. | `/mentor-me` | Think through a leadership situation with an Engineering Director mindset |
  80. | `/one-on-one` | Prepare for a 1:1 — situation read, key questions, risks, and opening line |
  81. | `/plan-initiative` | Structure a rough idea into a scoped initiative with epics and next steps |
  82. | `/us-mapping` | Generate a User Story Map from a PRD or Figma design |
  83. | `/write-epic-build` | Write a delivery epic with scope, risks, and definition of done |
  84. | `/write-epic-technical-discovery` | Write a discovery epic focused on reducing uncertainty and enabling a decision |
  85. | `/write-strategy` | Draft a strategy doc (Rumelt + Larson) with diagnosis, policies, and actions |
  86. | `/write-update` | Write a periodic update (weekly, monthly) by pulling Jira, Slack, and URL sources and formatting the output with a chosen template |
  87. | `/write-us` | Draft a user story following INVEST and vertical slicing |
  88. | `/write-vision` | Draft a vision doc with value proposition, capabilities, and constraints |
  89. ---
  90. ## Workspace structure
  91. ```
  92. claude-em/
  93. ├── data/ # Shared data across initiatives
  94. │ ├── em_style.md # Engineering Manager style
  95. │ ├── team_{name}.md # Team context files
  96. │ ├── [source]/ # One folder per data source (jira, github, etc.)
  97. │ └── tmp/ # Temporary files
  98. └── initiatives/ # All initiatives live here
  99. └── {initiative-name}/ # One folder per initiative
  100. ├── data/ # Initiative-specific data
  101. ├── tmp/ # Temporary files
  102. ├── scripts/ # Analysis and processing
  103. └── output/ # Results and reports
  104. ```
  105. ---
  106. ## Setup (full)
  107. ### 1. Create your workspace
  108. Click on **[Use this template](https://github.com/new?template_name=claude-em&template_owner=jcesarperez)** and create your own repository.
  109. This gives you a fully independent workspace you can customize.
  110. ---
  111. ### 2. Install Claude Code
  112. ```bash
  113. npm install -g @anthropic-ai/claude-code
  114. ```
  115. > Docs: https://docs.anthropic.com/claude-code
  116. ---
  117. ### 3. Configure your EM profile
  118. Define your EM style:
  119. * Copy `data/em_style_example.md` → `data/em_style.md`
  120. * Fill in your style and how you want Claude to respond
  121. Then create your team context:
  122. * Copy `data/team_example.md`
  123. * Rename it to `data/team_{your-team}.md`
  124. * Fill in your team context and members
  125. ---
  126. ### 4. (Optional) Install CLI tools
  127. Claude EM is designed to work primarily with CLI tools instead of MCP, as they are faster and more cost-efficient.
  128. **Jira CLI (`jira`)**
  129. ```bash
  130. brew install ankitpokhrel/jira-cli/jira
  131. jira init
  132. ```
  133. > Docs: https://github.com/ankitpokhrel/jira-cli
  134. **GitHub CLI (`gh`)**
  135. ```bash
  136. brew install gh
  137. gh auth login
  138. ```
  139. > Docs: https://cli.github.com
  140. ---
  141. ### 5. (Optional) Configure credentials
  142. Create a `.env.local` file at the workspace root (git-ignored):
  143. ```bash
  144. # Jira
  145. JIRA_URL="https://your-domain.atlassian.net"
  146. JIRA_EMAIL="your-email@company.com"
  147. JIRA_API_TOKEN="your-api-token"
  148. # GitHub
  149. GITHUB_TOKEN="your-github-token"
  150. # Figma
  151. FIGMA_TOKEN="your-figma-token"
  152. ```
  153. Getting credentials:
  154. - **Jira API token**: https://id.atlassian.com/manage-profile/security/api-tokens
  155. - **GitHub token**: https://github.com/settings/tokens (scopes: `repo`, `read:org`, `read:user`)
  156. - **Figma token**: https://www.figma.com/settings → Personal access tokens
  157. ---
  158. ### 6. (Optional) Configure the Jira skill for your project(s)
  159. The `/jira` skill works with any Jira project, but you need to point it at yours. When you invoke it for the first time, Claude will ask for the missing values and offer to save them permanently.
  160. If you manage **more than one project**, create one skill file per project:
  161. ```bash
  162. # Copy the base skill for each project
  163. cp .claude/skills/jira-project/SKILL.md .claude/skills/jira-myproject/SKILL.md
  164. cp .claude/skills/jira-project/SKILL.md .claude/skills/jira-otherproject/SKILL.md
  165. ```
  166. Then open each copy and update:
  167. 1. **Folder name** — already done by the `cp` command above
  168. 2. **`name:` in the frontmatter** — must match the folder name (e.g. `jira-myproject`)
  169. 3. **`description:` in the frontmatter** — replace "any Jira project" with your project name so Claude picks the right skill when you have multiple
  170. 4. **The three configuration placeholders:**
  171. | Placeholder | Example |
  172. |---|---|
  173. | `{{PROJECT_KEY}}` | `PLAT` |
  174. | `{{CLOUD_ID}}` | `abc123-...` (from `jira config list`) |
  175. | `{{BASE_URL}}` | `yourcompany.atlassian.net` |
  176. Each copy becomes its own skill. Claude will pick the right one based on context, or you can invoke them explicitly.
  177. ---
  178. ### 7. (Optional) Configure MCP servers
  179. MCP servers give Claude direct access to Jira, GitHub, and Figma without CLI or scripts.
  180. ```bash
  181. claude mcp add --plugin atlassian
  182. claude mcp add --plugin github
  183. claude mcp add --plugin figma
  184. ```
  185. **Slack (MCP only)**
  186. Slack has no practical CLI, so reading channels/threads is done through MCP. Add the official Slack connector and authenticate with your account:
  187. ```bash
  188. claude mcp add --transport http slack https://mcp.slack.com/mcp
  189. ```
  190. > Note: Slack access depends on your channel/workspace permissions. DMs are not readable, and the history tools ignore date filters — Claude pulls recent messages and filters the window itself.
  191. **Tip:** record your team's key Slack channels in `data/team_{name}.md` so Claude knows where to look (e.g. an `## Slack` section listing channel names). Otherwise it has to search and ask which channels apply.
  192. ---
  193. ### 8. (Optional) Get the most out of 1:1s
  194. The `/one-on-one` skill works with nothing but a one-line prompt, but it gets sharper the more context you give it through the team file.
  195. **Per-person 1:1 doc.** Give any member a living doc and the skill will read it when preparing the conversation — open topics, agreed actions, signals to watch, goals, and what happened in past 1:1s:
  196. * Add a `1:1 file:` link to the member in `data/team_{your-team}.md`
  197. * The value can be a URL or a path, and the doc can be in **any format** — the skill adapts to whatever's there
  198. * For a suggested starting point, see `.claude/skills/one-on-one/references/person-file-example.md`
  199. * It's fully optional: if it's missing or unreachable, the skill simply ignores it
  200. **Career Frameworks.** Fill in the `## Career Frameworks` section of the team file with links (per level) to your expectations docs. The skill reads the one matching the member's seniority when the conversation turns to expectations or growth.
  201. ---
  202. ### 9. (Optional) Configure `/write-update`
  203. The `/write-update` skill writes periodic team or project updates by pulling data from Jira, Slack, and configured URLs. It works out of the box with a one-line prompt, but gets significantly better with two things set up in the team file.
  204. **Templates.** The skill ships with two built-in templates in `.claude/skills/write-update/assets/`:
  205. | Template | Purpose |
  206. |---|---|
  207. | `weekly_team_update.md` | Weekly update for the whole org — highlights, lowlights, pipeline, metrics, allocation |
  208. | `weekly_project_update.md` | Status update for a specific project or initiative — status indicator, highlights, lowlights, upcoming |
  209. You can create your own templates by adding `.md` files to that same `assets/` folder. When you invoke the skill without specifying a template, Claude lists the available ones and asks you to choose. Each template can include an `## Instructions` section to tell the skill which sources to query and how to scope the data.
  210. **Update Sources.** Add an `## Update Sources` section to `data/team_{your-team}.md` listing the URLs Claude should read — Confluence meeting notes, Atlas project pages, and so on. The skill fetches each URL in parallel and extracts relevant content for the date range.
  211. **Slack channels.** Add a `## Slack` section to the team file with the channel names to include in each update. Claude reads them in parallel and filters for signal (shipped work, decisions, risks, customer feedback).
  212. ---
  213. ## Notes
  214. * You don’t need full setup to start — Quick Start is enough
  215. * CLI tools and MCP improve automation but are optional
  216. * The system works best when your team context is accurate and up to date
  217. ---
  218. ## Updating
  219. Your workspace is yours.
  220. Check the repository releases to see:
  221. - New skills
  222. - Improvements
  223. - Bug fixes
  224. Pull improvements when they’re useful.
  225. ---
  226. ## Contributing
  227. Contributions are welcome:
  228. * Feedback from real usage
  229. * Suggestions
  230. * New skills
  231. * Support for additional tools
  232. ---
  233. ## License
  234. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.