Engineering Manager

    Claude Code setup for engineering managers balancing people leadership, technical direction, and delivery execution. Covers 1:1s, sprint management, hiring, and team development.

    intermediate
    84 views
    engineering-managementone-on-onessprint-managementhiringteam-development

    CLAUDE.md

    CLAUDE.mdpaste in your project root
    # CLAUDE.md — Engineering Manager
    > Version: 2026-02-17 | Difficulty: intermediate
    
    ## What Your AI Can Now Do
    - Prepare for every 1:1 with a summary of each report's recent PRs, on-call shifts, and open feedback
    - Generate sprint retros pre-filled with velocity data and last retro's action items
    - Track career development conversations and surface growth themes over time
    - Flag blocked work items before they stall the sprint
    - Draft performance review summaries from months of 1:1 notes and feedback history
    
    ---
    
    ## Quick Start
    1. Add your team members to `team/members/` with their roles, strengths, and growth areas
    2. Run `/one-on-one` before your next 1:1 to see AI-powered prep in action
    3. Set up GitHub and calendar integrations in `.env` for delivery tracking
    4. Log your current sprint plan in `delivery/sprints/`
    5. Run `/team-health` to get your first biweekly team assessment
    
    ---
    
    ## About Me
    Engineering manager leading 6-10 engineers from junior through senior level.
    Time split: people management (40%), delivery (35%), technical direction (25%).
    I review code but don't write production code regularly. Running 2-week sprints.
    
    ---
    
    ## Pre-Built Commands
    
    | Command | What It Does |
    |---------|-------------|
    | `/one-on-one` | Prep for a 1:1 — recent work, open feedback, and suggested talking points |
    | `/sprint-retro` | Generate retro with velocity data and last retro's action item status |
    | `/team-health` | Biweekly assessment — velocity, review load balance, morale signals, risks |
    | `/perf-summary` | Draft performance review from 1:1 notes, feedback log, and delivery data |
    | `/hiring-plan` | Structure an open role with rubric, interview plan, and sourcing criteria |
    
    ---
    
    ## How Claude Connects to Your Tools
    
    | Tool | Why You Need It |
    |------|----------------|
    | Google Workspace | 1:1 docs, sprint notes, performance reviews, hiring coordination |
    | GitHub | PR velocity, code review load distribution, deployment tracking |
    | Memory | 1:1 themes over time, career development tracking, team patterns |
    | Supabase | Delivery metrics, cycle time data, deployment frequency |
    
    ### Setup
    ```bash
    # Required in .env
    GOOGLE_WORKSPACE_EMAIL=your-email@company.com
    GITHUB_TOKEN=your-token
    SUPABASE_URL=your-project-url
    ```
    
    ---
    
    ## People Leadership Rules
    - **MUST:** Never cancel a 1:1 — it's the most important meeting on your calendar
    - **MUST:** Deliver feedback within 48 hours using SBI (Situation, Behavior, Impact)
    - **MUST:** Protect 20% of sprint capacity for bugs, debt, and exploration
    - **SHOULD:** Hold quarterly career conversations separate from performance reviews
    - **MUST:** Performance reviews contain zero surprises — if they're surprised, you failed
    - **SHOULD:** Evaluate hires for culture ADD, not culture fit — rubric-based, decide in 48 hours
    - **MUST:** If you're thinking about firing, you're already 3 months late — document, coach, act
    - **SHOULD:** Ask "Is this a people problem or a process problem?" — fix the right one
    
    ---
    
    ## Project Structure
    ```
    team/
      members/
      one-on-ones/
      health/
    delivery/
      sprints/
      metrics/
      blockers/
    hiring/
      roles/
      pipeline/
      onboarding/
    development/
      growth-plans/
      feedback-log/
    ```
    
    ### What These Folders Mean
    - `team/` — Individual profiles, 1:1 notes by person, and biweekly health assessments
    - `delivery/` — Sprint plans with commitments vs. actuals, velocity trends, blocker log
    - `hiring/` — Open roles with rubrics, candidate pipeline, and 30-60-90 onboarding plans
    - `development/` — Growth plans with milestones and a log of all feedback delivered
    
    ---
    
    ## Automation
    
    | When | What Happens | If It Fails |
    |------|-------------|-------------|
    | 1:1 scheduled | Auto-pulls recent PRs, on-call shifts, and open feedback for that person | Review their GitHub activity manually before the meeting |
    | Sprint end | Generates retro template with velocity data and last retro's action items | Create blank retro doc and flag missing metrics |
    | Blocked item hits 24 hours | Escalation reminder with suggested owner and context | Add to next standup agenda as unresolved blocker |
    | Quarterly review cycle | Surfaces 1:1 themes, feedback history, and growth plan per person | Compile from 1:1 notes folder — flag anyone missing recent notes |
    | New hire starts | Generates 30-60-90 day plan from role rubric and team context | Use generic onboarding template and customize in first 1:1 |
    
    ---
    
    ## Gotchas
    - **MUST:** If something gets added mid-sprint, something else comes out — no exceptions
    - **MUST:** Fix code review bottlenecks as a team design problem — one reviewer is a bus factor risk
    - **SHOULD:** "I'm fine" in a 1:1 means trust isn't built yet — keep showing up consistently
    - **MUST:** Never promote your best IC into management without training — you'll lose two roles
    - **SHOULD:** Your team's output is your output — stop measuring yourself by personal contributions
    
    ---
    
    ## Memory & Learning
    Claude builds context over time:
    - 1:1 conversation themes and career development progress per person
    - Sprint velocity trends and what caused deviations
    - Feedback delivered and whether it led to observable behavior change
    - Team composition changes, onboarding status, and hiring pipeline
    - Retro action items and their completion rate over time
    
    To save something important: just say "remember that [fact]" in conversation.
    To review saved memories: use the `/memory` command to open the memory editor.
    Each setup follows the 5-layer Context Engineering framework: Identity, Memory, MCP, Skills, and Hooks.
    Learn the framework

    Pre-Built Skills

    Ready-to-use skills included in this setup. Each runs a specialized workflow for your role.

    📊/sprint-review

    Summarize sprint outcomes and velocity trends

    👤/1on1-prep

    Prep agenda for direct report check-ins

    🔄/retro

    Facilitate retrospective with structured prompts

    💚/team-health

    Assess team morale and capacity indicators

    Recommended MCP Servers

    Google Workspace MCP

    1:1 docs, sprint notes, and performance reviews

    Memory MCP

    Track 1:1 themes and career development progress

    Supabase MCP

    Query delivery metrics and deployment data

    Workflow Templates

    1:1 Preparation

    /one-on-one

    Prepare for 1:1 meetings with recent work review and discussion topics

    Sprint Retrospective

    /sprint-retro

    Facilitate sprint retro with structured reflection and action items

    Team Health Check

    /team-health

    Assess team health across delivery, morale, and risk indicators

    This Is a Starter Config

    MemoryOS analyzes YOUR actual workflow, tools, and communication style to generate a brain personalized to you. Fully automated — you decide what to share, it does the rest.

    Based on patterns from 100+ professionals. No technical knowledge required.

    Get Your Personalized Setup