How to Turn Claude Code into an Operational Partner with makerskills

When you're launching your own project or working on a solo startup, you end up being the developer, the marketer, and the CFO all at once. Most of the time, we use AI tools like Claude Code, Cursor, or Codex only for writing code. But what if you could make the agent handle operational tasks following strict methodologies?

The makerskills project author, Cory Haynes, has compiled a plugin with 19 ready-made scenarios for AI agent environments. These aren't generic prompt collections—they're structured protocols that transform an assistant into a systematic helper.

Documentation philosophy first and foremost

Most prompt libraries suffer from vague phrasing. Makerskills takes a different approach.

Each skill is documented in a separate file SKILL.md. It's a detailed step-by-step guide. You can invoke the skill via a console command (for example, /decide) or simply open the markdown file and follow the steps manually. Automation here turns out to be a side effect of good documentation.

The separation between public code and personal information is cleverly designed. The repository itself is fully open, while personal settings, workspace IDs, financial reports, and knowledge bases are stored locally on disk in ~/.config/makerskills/. Your secrets and internal notes won't leak into the public repository.

What these 19 skills can do

All scenarios are broken down into several categories for specific work tasks.

Meta-skills for expanding the system

The plugin can teach itself and expand the environment's capabilities.

• skillify creates, adapts, or updates skills right during a conversation or based on video recordings.
• toolify attaches a third-party API, integration, or MCP server to a Next.js or Rails project. The tool walks you through a step-by-step setup wizard for authentication, environment variables, and webhooks.
• loopify configures background loops and scheduled tasks for agents with stop condition checks.

Decision-making and strategy

• decide uses the 37signals decision-making framework with 38 questions. The agent selects 6-8 relevant questions for your situation, helps weigh risks, and saves the result with a review date.
• unstuck helps you get unstuck when a task seems impossible. The agent classifies the cause of the blocker and applies lateral thinking across 10 different angles before giving up.
• domain runs an 11-step search for available .com domains. The tool uses Vercel CLI, WHOIS, RDAP API, checks trademarks in USPTO via a headless browser, and suggests alternatives without being blindly attached to specific names.
• deep-research conducts deep research across multiple sources with citation formatting and result saving.

Knowledge and content work

• second-brain and company-brain organize a personal or team knowledge base on top of markdown files. Information is periodically cleaned of outdated details so the model doesn't give wrong answers based on stale protocols.
• read-book and watch-video extract key points, quotes, and summaries from books (PDF, EPUB) or video recordings (YouTube, Loom, Zoom).

Operations and utilities

• personal-cfo and company-cfo calculate financial scenarios, cash flow, and compile monthly company reports.
• paste cleans terminal output before sending to Slack, Notion, or social media, while simultaneously checking for accidental API key and password leaks.
• social-fetch pulls publications by URL from 10 social platforms as structured data.

How skills work together

The strong side of makerskills lies in synergy. Skills reference each other by name.

If you process a meeting recording through watch-video, the tool can send the summary to second-brain, flag tasks for pm, and log disputed decisions in decide.

The skillify skill can turn a recorded screen demonstration video into a new ready-to-execute scenario. Individual utilities stack into an end-to-end workflow.

Quick start

Installing the plugin in Claude Code takes a couple of minutes:

/plugin marketplace add coreyhaines31/makerskills
/plugin install makerskills@makerskills

Then set a base environment variable in ~/.zshenv:

export MAKERSKILLS_CONFIG="$HOME/.config/makerskills"

After restarting the shell, you can immediately try the first skill, which requires no prior setup:

/decide

The agent will ask which dilemma you're solving, select questions, and save the result to disk.

To connect a personal knowledge base or financial planner, you'll need to specify paths to your directory folders in the SECOND_BRAIN_VAULT or COMPANY_CFO_ROOT variables.

Who will find this repository useful

The makerskills project is built for those running projects solo or in small teams. It suits indie hackers, solo founders, tech leads, and freelancers.

If you regularly give AI similar complex tasks around research, note-taking, or financial modeling, ready-made scenarios will save hours of writing detailed instructions from scratch.