Aidan's Agent Skills
A collection of skills for Claude Code that give it new tricks for common dev work.
Getting started
Open Claude Code and run:
/plugin marketplace add awojtas/aidans-agent-skills
After that, open /marketplace again, navigate to "Aidan's Agent Skills", and pick which plugins you want to install.
Then close and re-open Claude Code - skills don't hot-reload, so you need a fresh session for them to show up.
How to contribute
Want to add a skill? Point your AI coding agent at this repo and tell it what you want to build. The AGENTS.md file has the full walkthrough for adding a new plugin and skill, so the agent can handle the scaffolding, file structure, and marketplace registration on its own.
Skills
Build Fixer (/build-fixer)
Runs your build, reads the errors, fixes them, runs it again. Keeps going until it compiles or it's clear something needs human attention.
Works with .NET, Node.js (npm/pnpm/yarn/bun), Rust, Go, Java (gradle/maven), Make, and others.
What it does:
- Figures out your build command from project files
- Classifies errors as fixable code problems vs. environment issues it can't solve
- Loops through fixes until the build passes or it stops making progress
- Tells you what's left if anything still fails
Ask Claude to fix build errors, clean up lint warnings, or just paste your build output.
Design System Aurora (/design-system-aurora)
A full design system built around glassmorphism, aurora gradients, neon glows, and a purple-cyan palette. Think frosted glass cards with glowing edges.
What it does:
- Gives you design tokens for colors, typography, spacing, and shadows
- Walks you through glassmorphic styling - backdrop blur, translucent borders, neon glow effects
- Includes aurora gradient animations and gradient text patterns
- Covers responsive design, dark mode, accessibility, and animation
- Comes with reference docs for tokens, glass patterns, and component examples
Use it when you're building or tweaking UI components and want them to match the Aurora look.
E2E Test Runner and Fixer (/e2e-test-runner-fixer)
Your E2E tests are failing. This skill figures out why and fixes them.
Works with Playwright, Cypress, and WebDriver.
What it does:
- Finds your runner config, test specs, helpers, and seed scripts
- Reproduces failures in isolation before touching anything
- Sorts failures into categories: real bug, flaky test, environment-dependent, platform-specific
- Digs into root causes like data setup mismatches, race conditions, stale locators, or auth leakage
- Makes the smallest fix that actually solves the problem, then re-runs to confirm
Use it when E2E tests break, specs are flaky, or your CI test job is red.
Issue Closer (/issue-closer)
Goes through your open GitHub issues and closes the ones where the work is already done.
What it does:
- Pulls up all open issues in the current repo
- Checks for evidence that they're finished - merged PRs, commits, code that matches the request
- Closes anything that's clearly done, with a comment explaining why
- Gives you a summary of what got closed and what's still open
Good for cleaning up a backlog that's gotten stale, or after a sprint where tickets didn't get closed along the way.
Issue Prioritiser (/issue-prioritiser)
Looks at your open GitHub issues and helps you figure out what to work on next.
What it does:
- Pulls issues matching a filter (milestone, label, or everything)
- Checks whether each issue is still relevant and well-defined enough to act on
- Rates priority based on impact, how often people hit the problem, and effort to fix
- Labels them: highest, high, medium, low, nice to have
- Flags issues that might already be done or too stale to matter
- Recommends a ranked list of what to tackle first
Use it when the backlog feels like a mess and you need a starting point.
Issue Worker (/issue-worker)
Give it a GitHub issue and it does the whole thing - reads the ticket, understands the codebase, writes code, adds tests, and makes sure everything passes.
What it does:
- Reads the issue and pulls out acceptance criteria and scope
- Explores the repo to understand conventions, build system, and stack
- Creates a branch and writes the minimum changes needed
- Adds unit and integration tests using whatever patterns already exist
- Runs lint, type-check, build, and tests, fixing anything that fails
- Double-checks against the acceptance criteria before reporting back
Say "work on issue #123" or drop in a GitHub issue link.
Sentry Recent Issues (/sentry-recent-issues)
Pulls recent Sentry issues, figures out what's going on in the code, and tells you how to fix them.
What it does: