# makerskills: a plugin of 21 agent skills for founders who run their own operations > makerskills packages decision frameworks, research workflows and knowledge-base routines as Agent Skills for Claude Code, Codex and Cursor. It is documentation-first, MIT licensed, and its last push was on 2026-09-04. **coreyhaines31/makerskills** — AI agent skills for the personal operator's craft — decisions, research, second-brain, content rotation, scenario modeling, and meta-skills to author more. Works with Claude Code, Codex, Cursor. - Repository: https://github.com/coreyhaines31/makerskills - Website: https://maker-skills.com - Stars: 839 · Forks: 69 - Language: Unknown - License: MIT - Published: 2026-09-10 · Updated: 2026-09-10 · Language: en - Canonical page: https://hysenlabs.com/projects/coreyhaines31-makerskills ## What makerskills is for, and who it is actually built for The README describes the project as "AI agent skills for the personal operator's craft" and names its audience directly: founders and indie operators. That framing matters, because the skill list is not a general-purpose toolset. It assumes one person making decisions, keeping notes, writing content across several projects and watching their own cash flow. The routing table in the README maps intents to skills. "Think through a decision with a real fork" goes to decide. "Break through a wall" goes to unstuck. "Model personal financial scenarios (house, cash flow, etc.)" goes to personal-cfo, while a company or agency gets company-cfo. There is a second-brain skill for a personal knowledge base and a company-brain skill for a team-shared one, which is the clearest signal of the intended scale: the same workflow, two different scopes. The problem it solves is not that agents cannot answer questions. It is that the operator's recurring work has no fixed shape, so every session starts from zero. makerskills answers that by writing the shape down as files you can read, edit and re-run. ## The mechanism: SKILL.md files, a plugin manifest, and a marketplace entry The README is explicit about where the logic lives: "Each skill is a workflow doc, you can invoke via `/decide` OR read the SKILL.md and run the steps by hand. The plugin is documentation-first; the automation is a side effect." That sentence should shape how you evaluate the project. A skill is a Markdown document under skills//SKILL.md, and the repository's top level carries the supporting scaffolding: .claude-plugin/ for the manifest, plus ARCHITECTURE.md, INSTALL.md, EXAMPLES.md and FAQ.md. Data flow is deliberately local. The README's description of the first-touch experience is "Structured input, structured output, written to disk." decide, for example, asks what decision you are facing, selects questions from a 37signals framework, and archives the result with a revisit date. The archive is a file, not a hosted record. Some skills reach outward. domain runs an 11-step hunt that the README says combines the Vercel CLI, whois per TLD, the Domainr API, the Namecheap API, rdap.org for modern TLDs, agent-browser for USPTO trademark screening, and aftermarket click-throughs to HugeDomains, Afternic, Sedo and Dan. deep-research lists WebSearch, WebFetch, agent-browser, /last30days and memory as its sources. That is a lot of surface area for a documentation-first plugin, and it is the main thing to inspect before you rely on those two skills. The meta-skills are the interesting design choice. skillify, toolify and loopify exist so you can author more skills in the same shape. The README calls them "the `-ify` trifecta" and gives skillify three modes: CREATE, ADAPT and UPDATE, where ADAPT is described as porting an external skill with a license check and attribution. A project that ships its own authoring path is betting that the 21 skills are a starting set rather than the product. ## Installing makerskills in Claude Code, and your first run of /decide The README gives two install routes. The first is the plugin marketplace inside Claude Code. The second is a symlink for local development, which is what you want if you intend to edit the SKILL.md files rather than only consume them. ```bash # In Claude Code /plugin marketplace add coreyhaines31/makerskills /plugin install makerskills@makerskills ``` For local work, the README clones the repository and links it into the plugins directory: ```bash git clone https://github.com/coreyhaines31/makerskills ~/code/makerskills ln -s ~/code/makerskills ~/.claude/plugins/makerskills ``` After installing, the README asks you to set one environment variable in ~/.zshenv and reload your shell: ```bash export MAKERSKILLS_CONFIG="$HOME/.config/makerskills" ``` The README points to INSTALL.md for environment variables, personal-config setup and runtime dependencies, so treat that file as the authoritative list rather than the README snippet. The first skill to run is decide, which the README says needs no config. You invoke it by typing the slash command in your agent session: ``` /decide ``` Expect it to ask what decision you are facing, then pick six to eight questions out of a larger set and archive the outcome with a revisit date. If nothing gets written to disk, the archive step is where to look first. The README then suggests a config-free utility as a second test, paste, invoked as `/paste twitter`, which reads your clipboard, strips formatting, warns if the result is over 280 characters and copies the cleaned text back. It notes that the first invocation may prompt you to install missing pbcopy-adjacent dependencies. That warning is worth taking literally: paste is the skill most likely to fail on a fresh machine for reasons that have nothing to do with the plugin. ## Where makerskills breaks down or is the wrong choice The README states that the repository is documentation-first, and that is also its main limitation. Nothing here is a service. If a skill's steps depend on an external tool you do not have, the skill does not degrade gracefully into something else; you either install the tool or run the steps by hand. domain is the clearest example. Its eleven steps lean on the Vercel CLI, whois, the Domainr API, the Namecheap API, rdap.org, agent-browser and several aftermarket sites. The README does not document what happens when one of those is unavailable or rate-limited, and it does not describe a fallback path. For a skill that ends in a purchase decision, that gap is worth knowing about before you start. There is a scope mismatch to watch as well. The README frames the project around a single operator, but several skills assume a business context: company-brain, company-cfo, ingest, pm. If you are a solo developer with no company entity, those skills have less to attach to. Finally, the plugin is not a substitute for the judgment it structures. decide triages a fixed question set and archives an outcome. It does not know your numbers. The archive is only as good as what you typed into it, and the revisit date only helps if something reads the archive later. ## How makerskills differs from a general Claude Code setup or a notes app The obvious alternative is running Claude Code with no plugin and writing your own prompt each time. The difference is not capability but persistence. A one-off prompt leaves no artifact; the makerskills pattern is structured input, structured output, written to disk. decide archives with a revisit date and second-brain captures and queries a knowledge base. If you never revisit your decisions, the plugin's advantage over ad-hoc prompting largely disappears. The second alternative is a dedicated notes or PKM application. Those give you storage, search and a stable interface. makerskills gives you a workflow document that an agent executes, which means the retrieval step can be conversational and the capture step can be triggered from a session. The trade-off runs the other way too: a notes app does not depend on an agent host being installed and configured, and it does not care whether your clipboard utility is present. The third point of comparison is the host itself. The README says makerskills works with Claude Code, Codex, Cursor "and other Agent Skills hosts." If you are already committed to one of those, the plugin is additive. If you are not, the plugin is not a reason to pick one. ## Maintenance, versioning and the MIT licence The repository is not archived, and its last push was on 2026-09-04. The release history shows steady, small increments rather than large drops: v1.3.0 added unstuck on 2026-07-15, v1.4.0 added the Q39 opportunity cost question to decide on 2026-07-16, and v1.5.0 on 2026-08-12 combined maker-council with a domain v0.2.0 bump. Two skills in the same release line carry their own version numbers, so upgrade cost is not uniform across the plugin. The README describes skillify's UPDATE mode as improving existing skills with "cross-skill propagation + semver discipline." That is the upgrade path to plan around: a change to a shared framework can ripple into several skills at once. If you have edited SKILL.md files locally through the symlink install, expect to reconcile those edits on each pull. The licence is MIT, stated in the README badge and present as a LICENSE file at the repository root. That permits commercial use and modification, and it requires the licence and copyright notice to travel with copies. The README also notes that skillify's ADAPT mode performs a licence check and attribution when porting an external skill, which is consistent with that. None of this is legal advice; read the LICENSE file and INSTALL.md before you redistribute anything. ## Conclusion Adopt makerskills if you already run decisions, research and bookkeeping by hand and want those routines written down as invocable documents inside Claude Code, Codex or Cursor. Do not adopt it if you want a managed service or a product with a support contract: it is a repository of SKILL.md workflows plus plugin metadata, and per INSTALL.md you are expected to configure environment variables and runtime dependencies yourself. Before committing, open skills/decide/SKILL.md and one config-dependent skill such as second-brain, and confirm every external tool they call is something you already have or are willing to install. ## FAQ ### What are maker skills? In this project, makerskills is a set of 21 Agent Skills aimed at the personal operator's craft: decisions, research, second-brain workflows, content rotation, scenario modeling and CFO cadence. The README describes them as workflow documents you can invoke with a slash command or run by hand. ### What is the best marketing skill in Claude? The README does not rank skills by quality or single out a marketing one. The closest fits in its routing table are jab-hook, which plans and drafts social content rotation across a portfolio, and paste, which cleans terminal output for Slack, LinkedIn, X and Notion. ### How do I build my own skills? The README points to reading an existing SKILL.md first, using skills/decide/SKILL.md as the pattern, then using skillify to author a new one. Each skill is a workflow document, so the steps can also be run by hand without the plugin. ### How do I create a skill? Use the skillify skill, which the README lists under the meta group with three modes: CREATE from chat, video, a dump or from scratch; ADAPT to port an external skill with a licence check and attribution; and UPDATE to improve existing skills with cross-skill propagation and semver discipline. ### What is maker skills? It is the same project. The README describes makerskills as AI agent skills for the personal operator's craft, built for founders and indie operators, and states that it works with Claude Code, Codex, Cursor and other Agent Skills hosts. ## Sources - [coreyhaines31/makerskills on GitHub](https://github.com/coreyhaines31/makerskills) - [License: MIT](https://github.com/coreyhaines31/makerskills/blob/main/LICENSE) - [Project website](https://maker-skills.com) - [README](https://github.com/coreyhaines31/makerskills/blob/main/README.md) - [Releases](https://github.com/coreyhaines31/makerskills/releases) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/coreyhaines31-makerskills