Skip to main content
A software engineering plugin designed for Cowork and Claude Code. Helps with standups, code review, architecture decisions, incident response, debugging, and technical documentation.

Installation

claude plugins add knowledge-work-plugins/engineering

Overview

The Engineering plugin transforms Claude into an engineering teammate that helps with common software development workflows. Every command works standalone (paste code, describe your system) and gets supercharged when you connect your development tools via MCP.

Commands

Explicit workflows you invoke with a slash command:

/standup

Generate a standup update from your recent activity — commits, PRs, tickets, and chat

/review

Review code changes for security, performance, style, and correctness

/debug

Structured debugging session — reproduce, isolate, diagnose, and fix

/architecture

Create or evaluate architecture decisions with ADR format and trade-off analysis

/incident

Run an incident response workflow — triage, communicate, mitigate, and write postmortem

/deploy-checklist

Pre-deployment checklist — verify tests, review changes, check dependencies, confirm rollback plan

/standup

Generate a standup update from your recent activity:
/standup
If your tools are connected, Claude will pull your recent commits, PR activity, and ticket updates. Otherwise, tell Claude what you worked on and it will format it for you.

/review

Review code changes with a structured lens on security, performance, correctness, and maintainability:
/review https://github.com/org/repo/pull/123
Share a PR link, paste a diff, or point to files. Get a structured review covering:
  • Security: SQL injection, XSS, auth bypass, secrets in code
  • Performance: N+1 queries, memory allocations, algorithmic complexity, missing indexes
  • Correctness: Edge cases, error handling, race conditions
  • Maintainability: Naming clarity, single responsibility, test coverage, documentation
Example output:
## Code Review: Add user authentication

### Summary
Adds JWT-based authentication to the API. Overall structure is good,
but there are critical security concerns that need addressing.

### Critical Issues
| # | File | Line | Issue | Severity |
|---|------|------|-------|----------|
| 1 | auth.ts | 42 | JWT secret hardcoded | 🔴 Critical |
| 2 | middleware.ts | 18 | No token expiration check | 🔴 Critical |

### Suggestions
| # | File | Line | Suggestion | Category |
|---|------|------|------------|----------|
| 1 | auth.ts | 67 | Consider rate limiting login attempts | Security |
| 2 | db.ts | 123 | N+1 query in user lookup | Performance |

### What Looks Good
- Comprehensive test coverage for happy paths
- Clear error messages for authentication failures
- Good separation of concerns between auth and business logic

### Verdict
Request Changes — Address critical security issues before merging

/debug

Walk through a structured debugging process:
/debug Users are getting 500 errors on the checkout page
Claude helps you systematically:
  1. Reproduce the issue
  2. Isolate the problem area
  3. Diagnose root cause
  4. Fix and verify

/architecture

Create or evaluate architecture decisions:
/architecture Should we use a message queue or direct API calls between services?
Get a structured ADR (Architecture Decision Record) with:
  • Options analysis
  • Trade-offs for each approach
  • Recommendation based on your context
  • Implementation considerations

/incident

Run an incident response workflow:
/incident The payments service is returning 503s
Claude helps you:
  • Triage severity
  • Draft status communications
  • Track timeline and actions
  • Generate postmortem when resolved

/deploy-checklist

Get a customized pre-deployment checklist:
/deploy-checklist auth-service v2.3.0
Claude creates a checklist based on your service and what’s changing:
  • Test verification
  • Dependency checks
  • Rollback plan confirmation
  • Breaking change review

Skills

Domain knowledge Claude uses automatically when relevant:
SkillDescriptionWhen It Activates
code-reviewReview code for bugs, security issues, performance, and maintainabilityWhen you share code and ask for feedback, or say “review this code”
incident-responseTriage and manage production incidents — status updates, runbooks, postmortemsWhen you mention “incident”, “outage”, or “production issue”
system-designDesign systems and services — architecture diagrams, API design, data modelingWhen discussing “architecture”, “system design”, or “how should we build”
tech-debtIdentify, categorize, and prioritize technical debt — build remediation planWhen you mention “tech debt”, “refactoring”, or “code quality”
testing-strategyDesign test strategies — unit, integration, e2e coverage, test plan creationWhen discussing “testing”, “test coverage”, or “how to test”
documentationWrite and maintain technical documentation — READMEs, API docs, runbooksWhen writing or updating documentation

MCP Integrations

Connect your tools for a richer experience:
CategoryExamplesWhat It Enables
Source controlGitHub, GitLabPR diffs, commit history, branch status
Project trackerLinear, Jira, AsanaTicket status, sprint data, assignments
MonitoringDatadog, New RelicLogs, metrics, alerts, dashboards
Incident managementPagerDuty, OpsgenieOn-call schedules, incident tracking, paging
ChatSlack, TeamsTeam discussions, standup channels
Knowledge baseNotion, ConfluenceADRs, runbooks, onboarding docs

Connector Placeholders

Plugin files use ~~category as a placeholder for whatever tool you connect in that category. For example:
  • ~~source control might mean GitHub, GitLab, or any other VCS with an MCP server
  • ~~project tracker might mean Linear, Jira, or Asana
  • ~~monitoring might mean Datadog, New Relic, or Grafana
Plugins are tool-agnostic — they describe workflows in terms of categories rather than specific products.

Standalone vs Supercharged

Every command and skill works without any integrations:
What You Can DoStandaloneSupercharged With
Standup updatesDescribe your workSource control, Project tracker, Chat
Code reviewPaste diff or codeSource control (pull PRs automatically)
Debug sessionsDescribe the problemMonitoring (pull logs and metrics)
Architecture decisionsDescribe the systemKnowledge base (find prior ADRs)
Incident responseDescribe the incidentMonitoring, Incident management, Chat
Deploy checklistsDescribe the deployCI/CD, Source control

Configuration

Create a local settings file at engineering/.claude/settings.local.json to personalize:
{
  "name": "Your Name",
  "title": "Software Engineer",
  "team": "Your Team",
  "company": "Your Company",
  "techStack": ["Python", "TypeScript", "PostgreSQL", "AWS"],
  "defaultBranch": "main",
  "deployProcess": "canary"
}
The plugin will ask you for this information interactively if it’s not configured.

Example Workflows

Morning Standup

/standup
If your tools are connected, Claude pulls your recent commits, PR activity, and ticket updates. Otherwise, tell Claude what you worked on: You: Finished the user auth refactor, reviewed two PRs, and started on the rate limiting feature Claude: Here’s your standup update:
### Yesterday
- ✅ Completed user authentication refactoring
- 👀 Reviewed PRs #234 and #235
- 🚧 Started implementation of rate limiting feature

### Today
- Continue rate limiting implementation
- Deploy auth changes to staging

### Blockers
- None

Code Review Session

/review https://github.com/org/repo/pull/123
Claude pulls the PR, analyzes the changes, and provides structured feedback on security, performance, and maintainability.

Debugging Production Issue

/debug API returning 500s for /checkout endpoint
Claude: Let’s debug this systematically. First, let me help you reproduce and isolate the issue. Reproduction:
  • When did this start?
  • Is it affecting all users or specific segments?
  • What’s the error rate?
[Interactive debugging session continues]

Data Plugin

SQL queries, data exploration, and analytics

Design Plugin

Design critique, system management, and UX writing

Build docs developers (and LLMs) love