claude-skills-marketplace

Why use it

Key features

Engineering Workflow plugin: git, test-fixing, implementing review feedback
Visual Documentation plugin: HTML pages with SVG diagrams, flowcharts, dashboards, timelines
Productivity plugin: code audits, codebase docs, usage analysis, project setup
Code Operations plugin: high-precision line-based insertion and bulk refactors
Execution Runtime for 90-99% token reduction on bulk operations

Live Demo

What it looks like in practice

claude-skills-marketplace-skill.replay ▶ ready

0/0

Install

Author repository · Claude Skills documentation

Use Cases

Real-world ways to use claude-skills-marketplace

Run a bulk refactor across 200 files without burning tokens

👤 Engineers doing codebase-wide changes ⏱ ~30 min intermediate

When to use: The same transformation needs to apply everywhere and the LLM would churn through too many tokens doing it file-by-file.

Flow

Define the transformation

Use the Code Operations plugin. Replace every import { foo } from 'old-lib' with the equivalent from 'new-lib'. Run via Execution Runtime.✓ Copied

→ Plugin generates a Python script that runs locally, not a per-file LLM pass
Review the diff

Show me the full diff grouped by type of change.✓ Copied

→ Structured diff you can skim

Outcome: Massive refactor done cheaply and reviewably.

Pitfalls

Regex-based transform breaks edge cases — Runtime supports AST-level transforms — use those for syntax-sensitive changes

Combine with: git · github

Implement every actionable comment on a PR

👤 Devs with a PR under review ⏱ ~20 min intermediate

When to use: Reviewer left 12 comments and half are straightforward mechanical changes.

Flow

Pull and triage

Use the Engineering Workflow plugin. Pull PR comments on this branch, group into mechanical vs needs-thought.✓ Copied

→ Two buckets with count per category
Auto-apply mechanical

Apply the mechanical ones as a commit.✓ Copied

→ Single commit addressing rote feedback

Outcome: Fewer review cycles by handling mechanical work in one commit.

Pitfalls

Misclassifying a subtle comment as mechanical — Default to 'needs-thought' when unsure; reviewer won't be annoyed by caution

Combine with: github

Generate visual architecture documentation for a service

👤 Engineers writing architecture docs ⏱ ~30 min intermediate

When to use: You need a doc with real diagrams, not just prose.

Flow

Run the Visual Documentation plugin

Use Visual Documentation plugin. Generate an architecture HTML page for the orders service with SVG flowcharts.✓ Copied

→ Self-contained HTML with SVG, navigable

Outcome: A polished architecture doc you can publish internally.

Pitfalls

Diagrams too complex to read — Ask for multiple smaller diagrams per concern, not one mega-chart

Combinations

Pair with other MCPs for X10 leverage

claude-skills-marketplace-skill + git

Workflow plugin drives the git flow end-to-end

Use the Engineering Workflow plugin to make the change, commit with a good message, push, open PR.✓ Copied

claude-skills-marketplace-skill + github

GitHub MCP for the remote side, this skill for the local ops

Apply the bulk refactor locally via this skill, then use GitHub MCP to open the PR.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
Engineering Workflow plugin	git/test/review task	Daily engineering tasks	0
Visual Documentation plugin	subject to document	Architecture / project docs	0
Productivity Skills plugin	codebase path	Project health checks	0
Code Operations plugin	transformation spec	Bulk refactors and insertions	Local CPU, ~0 tokens
Execution Runtime	Python operation script	Under the hood for any bulk op	0 tokens for the bulk work

Cost & Limits

What this costs to run

API quota: None
Tokens per call: Very low for bulk ops thanks to the Execution Runtime
Monetary: Free — Apache 2.0, skill is local
Tip: Push as much as possible through the Execution Runtime; only use token-heavy paths when AST-level intent is needed.

Security

Permissions, secrets, blast radius

Credential storage: No credentials — skill is local plugins

Data egress: None from the skill itself

Execution Runtime runs generated Python with your local privileges — review the generated script before running on important code.

Troubleshooting

Common errors and fixes

Execution Runtime script errors on edge files

The plugin should skip and report; review the skip list before considering the run complete.

Visual doc SVG is broken in some browsers

Stick to self-contained SVG without external font references.

PR-comment implementation applied wrong change

Always review the commit before pushing; skill is helper, not autopilot.

Alternatives