How can we help you?

ProfessorGPT is a platform for creating, discovering, and distributing AI skills — structured Markdown files (SKILL.md) that define expert behaviors for Claude Code and other Claude-based agents. Think of it as a package manager for AI behavior: instead of re-writing prompts from scratch, you browse or generate a skill, publish it to the shared registry, and install it in your Claude Code sessions in seconds.

A SKILL.md file is a structured Markdown document with a YAML frontmatter header (name, version, description, category, tags, allowed-tools) followed by natural-language instructions that define an AI's role, step-by-step behavior, output format, and constraints. Skills are designed for Claude Code — once installed, they appear as slash commands in your sessions and immediately shape how the AI responds in that context.

There are three ways: (1) Browse the skill library and copy a ready-made skill as a starting point. (2) Use the Create page — describe what you want and our AI generates a complete SKILL.md in seconds. (3) Use Skill Builder Pro (available on paid plans) for a guided, multi-step generation workflow with advanced controls. Once created, you can publish your skill to the community registry or install it directly in Claude Code.

Browsing and installing skills from the library is always free without an account. The Free plan gives registered users 1 AI-generated skill creation per month using the basic AI model — enough to try the platform and get started. Paid plans (Pro, Studio, Enterprise) increase your monthly generation quota and unlock more powerful AI models for higher-quality output, plus features like Skill Builder Pro and private skills.

No account is required to browse and install skills from the library. Creating a free account is required to generate new skills with AI, save skills to your personal library, publish skills to the community, and track your usage. Paid plans additionally unlock Skill Builder, private skill storage, higher monthly generation quotas, and access to more powerful AI models.

When you fill out the Create form and hit Generate, your inputs are sent to our AI model. The AI analyzes your description, goals, target behavior, and any example interactions you provide, then generates a fully structured SKILL.md following best practices — correct frontmatter, clear role definition, step-by-step instructions, output format, and constraints. Generation typically takes 5–15 seconds. You can then edit the result inline before saving or publishing.

The standard Create page is a single-step generator: describe → generate → done. Skill Builder Pro (available on paid plans) is a guided multi-step workflow that walks you through defining the role, instructions, output format, constraints, and test cases one section at a time. It also supports iterative refinement (regenerate only a section), version history, and richer metadata. Skill Builder is ideal when you need precise, production-grade skills.

Absolutely. After generation the skill opens in a live Markdown editor where you can freely modify any section. You can tweak the role description, add domain-specific constraints, adjust the output format, or add example interactions. Changes are reflected in real time. When you're happy, save the skill to your library or publish it to the community registry.

ProfessorGPT has two skill sources: (1) Official skills — created and maintained by the ProfessorGPT team, covering common productivity, coding, writing, and analysis use cases. These are quality-reviewed and kept up to date. (2) Community skills — generated and submitted by registered users through the platform, then merged into the agent-skills GitHub repository after automated validation. Both types are free to browse, install, and use.

Once you've created and refined a skill, go to Dashboard → My Skills, open the skill, and click Publish. The platform automatically creates a GitHub Pull Request in the agent-skills repository on your behalf, runs automated validation (frontmatter format, security scan, structure check), and — once checks pass — merges it and updates the public registry. Your skill then appears in the Browse library for everyone to install.

The version field (e.g. 1.0.0) lets you track iterations of a skill over time. When you refine a skill after testing — whether changing the instructions, adding constraints, or updating the output format — increment the version number. We follow semantic versioning: bump the major version (1→2) for breaking changes in behavior, minor (1.0→1.1) for new features, patch (1.0.0→1.0.1) for small fixes.

SKILL.md files are purpose-built for Claude Code — Anthropic's official AI coding assistant. Once installed, skills appear as slash commands in your Claude Code sessions. Because the SKILL.md body is plain Markdown, you can also copy it manually and paste it as a system prompt in Claude Projects, or any other platform that accepts custom instructions. The CLI-based install workflow, however, is specific to Claude Code.

Skills in Claude Code are installed via the CLI: run npx skills add <owner/skill-name> in your terminal. For example: npx skills add professor-gpt/code-reviewer. This downloads the SKILL.md file into your project's .claude/skills/ directory. Claude Code automatically discovers installed skills — invoke them with /skill-name in any session. Add the -g flag to install globally across all your projects.

Yes, with a manual step. Since SKILL.md files are plain Markdown, you can copy the skill content and paste it as a system prompt in Claude Projects, ChatGPT Custom GPTs, or any API-based platform that accepts custom instructions. The CLI install workflow and automatic slash-command integration are specific to Claude Code, but the underlying skill content works anywhere a system prompt is accepted.

Use the skills CLI: run npx skills add <owner/skill-name> in your terminal. For example: npx skills add professor-gpt/code-reviewer. Add -g to install globally across all your projects. Run npx skills list to see installed skills, npx skills update to pull newer versions, and npx skills remove <skill-name> to uninstall. The CLI requires Node.js 18 or later.

Free plan: browse and install all community skills, 1 AI-generated skill creation per month using the basic AI model, basic editor. Pro & Studio plans: higher monthly skill generation quota, enhanced AI model for better-quality output, Skill Builder Pro multi-step workflow, and private skill storage. Enterprise plan: highest quota, best-in-class AI models, and team collaboration features. See the Pricing page for exact details on each plan.

We accept all major credit and debit cards (Visa, Mastercard, Amex) via LemonSqueezy. Monthly and annual billing options are available — annual plans offer a discount. You can manage your subscription, update payment details, and view invoices from the Dashboard → Billing page.

Go to Dashboard → Billing and click Cancel Subscription. Your paid plan remains active until the end of the current billing period. After cancellation you return to the Free plan — your saved skills and generation history remain accessible.

Prompts you send to the Create page are forwarded to our AI model to generate your skill. We do not store your generation inputs long-term. Skills you explicitly save to your library are stored in your account. We never use your private skills or generation inputs to train models or share them with third parties.

agent-skills (github.com/professor-gpt/agent-skill) is the open-source registry that powers the skill library. It contains all community and official skills as SKILL.md files, a generated registry.json for fast API lookups, and CI/CD workflows that validate new submissions and auto-deploy updates to professorgpt.ai. Skills are published through the ProfessorGPT platform — when you click Publish, the platform opens the Pull Request in this repository automatically on your behalf.

Test Suite (coming soon) is ProfessorGPT's automated evaluation layer for skills. It lets you define test cases — expected inputs and outputs — and run them against any SKILL.md to measure quality, catch regressions, and compare versions side-by-side. It also integrates with GitHub Actions so you can fail PRs that break quality thresholds before they reach production.

SkillOps (coming soon) is ProfessorGPT's collaboration layer for teams. It adds a centralized skill registry with version control and approval workflows, team permissions and role-based access, usage analytics (which skills are used, by whom, how often), deployment pipelines (dev → staging → production), and enterprise SSO via Okta, Azure AD, or Google Workspace.

Yes. The specification lives in the agent-skills repository at SPEC.md. Required frontmatter fields: name (kebab-case slug), version (semver), description (1–2 sentences). Optional fields: category, tags (array), author, allowed-tools (array of tool identifiers). The body after the closing --- is free-form Markdown. Validation is enforced by the CI workflow on every Pull Request.

The allowed-tools field tells Claude Code which tool permissions to automatically enable when a skill is loaded. For example, a skill with allowed-tools: [Read, Edit, Bash] tells the runtime that this skill needs file reading, editing, and shell access. In Claude Code, this maps to the --allowedTools flag. Tools not in the list remain restricted, keeping skill behavior predictable and safe.