Consulting

Claude skills built for real work.

We build custom Claude Code skills — reusable AI workflows that turn complex, multi-step tasks into single commands. From full-site SEO audits to video editing by conversation.

Book a Call

Claude Code is Anthropic's agentic coding tool — it reads files, runs commands, calls APIs, and writes code. A skill is a reusable definition that tells Claude exactly how to handle a specific type of work: what tools to use, what agents to spawn, what output to produce, and what quality gates to enforce.

We've built over two dozen production skills since the platform launched. Every skill listed below was written by our team, is used in daily operations, and is actively maintained. We also build custom skills for clients — if you have a repeatable workflow that involves research, analysis, content, or data, it's probably a good fit.

Media & Content

Photo Editor
Layered photo editing via conversation: ImageMagick for resize, crop, convert, and sharpen. AI background removal via rembg. Gemini-powered complex edits including object removal, inpainting, and restoration.
TikTok Transcriber
Transcribes TikTok videos (or any local audio/video) into clean Markdown: one-sentence executive summary, cleaned-up transcript, and source attribution. Handles multiple URLs in batch.
Meeting Notes Summarizer
Summarizes meeting transcripts into structured CEO-level notes with executive summary, key decisions, action items, and follow-ups. Pulls from Otter.ai or raw transcript files.

Web & WordPress

Static Site Publisher
Bootstraps and publishes static websites on Google Cloud Storage with Cloudflare CDN. Two modes: new site creation with DNS and SSL setup, or content updates to existing sites.
WooCommerce Diagnostic
Runs a fixed-fee, fixed-scope diagnostic of WooCommerce sites with custom ERP integrations. Produces a structured findings report with prioritized remediation steps.
WordPress Column Poster
Batch-posts newspaper columns to WordPress as drafts via the WP REST API. Handles formatting, categorization, and scheduling for recurring content publishers.
Competitor Site Evaluation
Runs a structured competitor website evaluation combining UX and messaging review, SEO metrics, content analysis, and strategic recommendations into a single deliverable.

Business Operations & Reporting

ADHD Triage
An ADHD-aware intake and review system. Captures new ideas with zero friction, enforces a 48-hour cool-down period before action, and runs weekly triage with a work-in-progress cap. Prevents shiny-object overload without losing good ideas.
KB Scribe
Reconstructs cross-surface activity and appends structured entries to the internal knowledge base. Queries Teamwork, Gmail, Google Calendar, and conversation history to build a complete log of what happened across all Claude surfaces since the last entry.
End-of-Day Summary
Generates a daily summary of all work completed across Claude surfaces — chat, Claude Code CLI, and Cowork. Logs what was done, what changed, and what's next.
Morning Briefing
Daily morning briefing that checks Gmail and top news headlines from the last 24 hours. Surfaces what needs attention before the workday starts.
Product Launch
Takes a Romeii service offering from idea to manual launch to productized SaaS. Structures the go-to-market process with phased milestones, pricing validation, and launch checklists.
Product Opportunity Scanner
Daily scan of Reddit, Product Hunt, Hacker News, SomebodyMakeThis.org, and PainBase.space for product opportunities that match Romeii's capabilities. Filters signal from noise.
Cowork Project Instructions
Generates consistent, well-structured instructions for new Claude Cowork projects. Ensures every project starts with clear scope, constraints, and deliverables.

Client-Specific Skills

We also build skills tailored to specific client workflows. These examples show the kinds of business-specific automation that skills make possible.

SEO Monitoring & Reporting
Automated SEO performance monitoring with weekly and monthly reporting. Pulls data from Search Console and analytics platforms into formatted deliverables.
Quarterly Business Reviews
Generates QBR presentations from accumulated weekly reports. Pulls metrics, highlights trends, and produces a ready-to-present deck with charts and recommendations.
Morning Dispatchers
Automated daily monitoring that runs via launchd or cron. Checks dashboards, flags anomalies, and sends alerts before the team arrives — no human trigger needed.
AI Prompt Management
Manages AI search prompts for tracking brand visibility in AI answer engines. Maintains prompt libraries and monitors ranking changes across LLM platforms.

Why skills instead of one-off prompts?

A prompt disappears when the conversation ends. A skill persists. It encodes the right tool calls, the right quality checks, the right output format — every time. Skills can orchestrate multiple subagents in parallel, connect to live data sources via MCP, and enforce consistency across runs. The difference is the same as between typing a command once and writing a script you can run forever.

How It Works

01

We scope the workflow.

Identify the repetitive or complex task, define inputs and outputs, and map the decision logic.

02

We build and test the skill.

Write the skill definition, wire up any MCP tools or APIs, and validate against real data.

03

We hand it over ready to run.

Install in your Claude Code environment with documentation. You own the skill file outright.

Frequently Asked Questions

A Claude skill is a reusable prompt-and-tool definition that runs inside Claude Code (Anthropic's CLI agent). Skills turn complex, multi-step workflows into single slash commands. They can orchestrate subagents, call MCP servers, read and write files, and produce structured output — all from a markdown definition file.

Yes. Skills run inside Claude Code (the CLI or desktop app). They are not standalone programs. You need an active Anthropic API plan or Claude Pro/Max subscription with Claude Code access.

Absolutely. Most of the skills listed here started as custom solutions for specific client problems. If you have a repeatable workflow that involves research, analysis, content generation, or data processing, it is likely a good candidate for a skill.

Custom skills are scoped and quoted per project after an initial discussion. Simple single-agent skills typically take a few hours. Multi-agent orchestration skills with MCP integrations take longer. You receive a fixed quote before any work begins.

MCP (Model Context Protocol) is an open standard that lets Claude connect to external tools and data sources — Google Search Console, DataForSEO, Slack, Google Drive, browser automation, and more. Skills can call MCP tools to pull live data into their workflows.

Need a custom skill built?

Book a Call
Book a Call