Plugins
Research, code-audit and reporting skills for Claude Code and Claude Cowork, packaged as installable plugins. Add the marketplace once, then install what you need.
Plugins run in Claude Code, the terminal agent (CLI). Adding a marketplace registers the catalogue; installing a plugin pulls in its skills, agents and commands.
Add the marketplace
/plugin marketplace add galileo14/galileo14-pluginsInstall a plugin
/plugin install <plugin>@galileo14-pluginsReplace <plugin> with research, ai or audit. Each group below carries its own command, ready to copy.
Browse and manage
Open the plugin manager to discover, enable, disable or update plugins from any of your marketplaces.
/pluginActivate your plugins
Apply install, enable and disable changes without restarting Claude Code.
/reload-pluginsOpen Personal plugins
In the Claude Desktop app, go to Customize → Personal plugins.
Add the marketplace
Click Create plugin → Add marketplace and paste the repository URL.
https://github.com/galileo14/galileo14-plugins
Install the plugins
Pick the plugins you want from the marketplace and install them. Their skills become available right away.
The Claude Code skills each plugin installs, what they do and what they are for.
Showing 13 of 13
No skill matches.
Try another search or clear the filter.
Research and intelligence gathering, from quick searches to full company dossiers.
/plugin install research@galileo14-pluginsA complete company dossier, as branded HTML.
Fans out 11 parallel researchers (own website, Google, LinkedIn, Reddit, Trustpilot, Google Maps, Glassdoor, Indeed, YouTube, press and Crunchbase) and synthesizes the findings into an HTML report. Before delivery, two independent graders verify claims and URLs.
Briefing yourself before a call or negotiation with a company's full public footprint, verified.
Exhaustive research on a topic via 50 parallel searches.
Splits the topic across 10 researchers running 5 searches each and consolidates everything into a single sourced markdown document. All researchers run in parallel, so the report lands in minutes.
Compiling everything published on a topic into one report without spending an afternoon searching.
Verifies a claim from 10 evidence angles.
Runs 10 parallel searches from distinct angles (primary sources, established media, fact-checkers, counter-evidence) and issues a structured verdict: true, false, misleading, partially true or insufficient data.
Confirming or debunking a claim before publishing it or repeating it in a meeting.
A quick panorama of any topic in five searches.
Runs 5 parallel web searches and returns a ~250-word summary directly in the conversation, no files generated.
Getting oriented on a topic in a couple of minutes before deciding whether it deserves deep research.
Meta-tooling for working with Claude: prompt engineering and skill scaffolding.
/plugin install ai@galileo14-pluginsTurns any request into a structured, ready-to-use prompt.
Applies one of three prompt-engineering frameworks (CAR, COSTAR or RISEN) depending on the task and returns a polished prompt to paste into any LLM. It asks which framework to use unless the request already names one.
Getting a professional prompt out of a rough idea without knowing the prompt-engineering frameworks.
Scaffolds a Claude Code skill that does not drift between runs.
Builds the skill on the five-controls deterministic framework (input, process, output, speed and tests), so it produces the same shape of result on every run.
Turning a prompt that works sometimes into a reliable, repeatable skill.
Codebase audits through four lenses (security, scalability, architecture and clean code), plus a Claude Code skill auditor.
/plugin install audit@galileo14-pluginsEvery codebase audit in a single pass.
Runs the security, scalability, architecture, clean-code and database lenses over the same project and consolidates the findings into a single severity-ranked report.
Getting the full picture of a project's state with one command.
Security audit of a codebase.
A read-only agent reviews the code with the OWASP Top 10 as its spine (confidentiality, integrity, availability and auditability) and delivers a severity-ranked report with file:line citations.
Knowing a project's security holes before deploying or inheriting it.
Audit of a project's growth ceilings.
Reviews database patterns, caching, concurrency, memory, I/O and background jobs, and delivers a ranked report of what will break as the project grows.
Knowing where the next bottleneck sits before traffic finds it.
Architecture audit: coupling, layering and debt.
Evaluates cohesion, coupling, abstraction quality, testability and tech debt, and delivers a severity-ranked report on the cost of changing the code.
Measuring how expensive a codebase is to evolve before committing to it.
Craftsmanship audit, file by file.
Reviews naming, function length, nesting, duplication, dead code and magic constants across the whole codebase, with file:line citations and fix steps. Not a diff review: it audits the entire project.
Finding a project's worst maintainability spots and where to start cleaning them.
Audit of database design and usage.
Reviews the project's schema, indexes, queries, migrations and data-access patterns, and delivers a severity-ranked report with file:line citations.
Catching N+1 queries, missing indexes and fragile schemas before they degrade the product.
Audits a Claude Code skill against the five-controls framework.
Checks whether each control (input, process, output, speed and tests) is present, wired and actually needed, flags over-engineering and reviews Claude Code best practices, with concrete improvements.
Knowing whether a skill is well built and what it lacks before handing it to a team.
We use analytics cookies to understand how the site is used. Nothing is sent until you accept. You can read the cookie policy.