workbench-dev

A developer toolkit for Claude Code that eliminates context switching in two areas where it hurts most: infrastructure operations and roadmap refinement.

The problem

During incidents

You get an alert. You open a terminal for kubectl get pods, another for kubectl logs, another for the RabbitMQ dashboard, another for psql... Five terminals, fifteen minutes, and you still don't know what's wrong.

During planning

A ticket lands in sprint. You open Jira, then Figma, then Confluence, then the codebase... Four tools, twenty minutes, and you still can't tell if it's ready to implement.

The solution

Define your stack once. Talk to Claude. Get answers.

You:    "check if my-api is running in dev"
Claude: 2/2 pods healthy, 0 restarts, 3m CPU, 86Mi memory

You:    "any errors in the logs?"
Claude: 12 occurrences of "column updated_at does not exist"

You:    "run migrations"
Claude: 4 pending migrations. Apply them? [confirm] → Done. 0 errors.

You:    "is PROJ-123 ready to implement?"
Claude: 3 blockers found — missing acceptance criteria, design doesn't match
        data model, no error states defined in Figma

Plugins

ops-suite — Infrastructure Operations

Manage services, logs, databases, queues, and deployments without leaving your terminal.

Skill	What it does	Auto-invocable?
service-status	Pod/container health, restarts, CPU, memory	Yes
service-logs	Error search, classification, pattern detection	Yes
port-forward	Local tunnels to cluster services	Yes
db-query	Natural language → SQL, read-only by default	Yes
queue-status	Queue listing, DLQ counts, consumer health	Yes
queue-triage	DLQ failure diagnosis, root cause analysis	Yes
deploy	Deploy merged PRs, verify health post-deploy	No — destructive
db-migrate	Run pending migrations with confirmation	No — destructive
queue-reprocess	Move DLQ messages back after fix	No — destructive

Supported stacks: Kubernetes, Docker Compose, ECS · RabbitMQ, Azure Service Bus, SQS, Kafka · PostgreSQL, MySQL · GitHub Actions, GitLab CI · MikroORM, TypeORM

Full documentation →

refinery — Roadmap Refinement

Analyze tickets, designs, documentation, and code to answer: "Is this ready to implement?"

Skill	What it does
clarify-ticket	Find ambiguities and missing info in tickets
ticket-analysis	Comprehensive cross-source analysis (ticket + design + docs + code)
sprint-review	Sprint readiness evaluation with 8 automated checks
analyze-design	Extract data fields, UI states, and actions from designs
analyze-docs	Search documentation across Confluence/Notion
analyze-data-model	Trace entity fields through the codebase
analyze-feasibility	Evaluate technical approaches against the codebase
create-bug	Structured bug reports with root cause analysis
adr	Architecture Decision Records
board-to-stories	Extract user stories from Miro/Excalidraw boards
notify-team	Draft and send team messages via Slack/Teams

Supported tools: Jira, Linear, GitHub Issues · Figma, Penpot · Confluence, Notion · Miro, Excalidraw · Slack, Teams, Discord

Full documentation →

creating-skills — Skill Authoring Guide

Meta-skill that teaches you how to create your own Claude Code skills following Anthropic best practices. Progressive disclosure, proper triggering, adapter patterns — all covered.

/creating-skills:creating-skills

How it works

All plugins share one design principle: config + adapters.

┌─────────────┐     ┌──────────────┐     ┌────────────────────┐
│ config.yaml │────▶│   SKILL.md   │────▶│ adapters/<tech>.md  │
│             │     │              │     │                    │
│ orchestrator│     │ reads config │     │ kubernetes.md      │
│ = kubernetes│     │ loads right  │     │ docker-compose.md  │
│             │     │ adapter      │     │ ecs.md             │
└─────────────┘     └──────────────┘     └────────────────────┘

1. You define your stack once in config.yaml (Kubernetes? RabbitMQ? PostgreSQL?)
2. Each skill reads your config and loads the right adapter
3. Adapters contain the real commands (kubectl, rabbitmqctl, psql) with {config.X.Y} placeholders

To add support for a new technology, create a markdown file. No code changes needed.

Installation

# From the Claude Code marketplace
/plugin marketplace add github:weorbitant/workbench-dev

# Install what you need
/plugin install ops-suite
/plugin install refinery
/plugin install creating-skills

Configuration

Each plugin has a config.example.yaml. Copy and fill in your values: