Guide · Agent skill
How to use the Rex Requirements Analyst agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
Rex is an agent skill: a folder of instructions that teaches an AI agent a specific way of working. This one makes the agent act as a requirements analyst, turning a fuzzy idea into a structured written spec that developers or other agents can build from without guessing. This guide explains what the skill does, how to install it, how to trigger it, and what to watch for, so you can get a usable requirements report on your first try.
Rex Requirements Analystsickn33/agentic-awesome-skills · MIT Download the skill Quick start
- 1Download the ZIP. Download the Rex skill ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into your project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Describe your idea. Write the feature or problem in plain language, including who it is for and what success looks like.
- 5Answer the clarifying questions. Reply to the up to three focused questions Rex asks, or skip them if your request already covers the details.
- 6Review the REX REPORT. Check the summary, MoSCoW list, user stories, edge cases and open questions before passing it on.
What the Rex skill makes the agent do
Given a fuzzy idea or feature request, Rex asks a few focused clarifying questions, separates must-have from nice-to-have features, writes user stories with testable acceptance criteria, and flags edge cases, security-sensitive areas and compliance concerns. Its output is a structured written report rather than code.
Rex is built to avoid implementation. The skill says it does not write code, design schemas or suggest implementations, so the report stays focused on what should be built and why.
- Asks at most three clarifying questions per round
- Sorts requirements using MoSCoW framing (Must Have, Should Have, Nice to Have)
- Writes user stories in the form: As a [role], I want [action] so that [outcome]
- Adds Given/When/Then acceptance criteria to each story
- Lists failure modes, boundary conditions, security-sensitive surfaces and performance-sensitive paths
- Lists open questions and marks whether each one is blocking
Install the Rex skill from the ZIP
Download the ZIP from this page. The skill is a single file, SKILL.md. To install it for yourself across projects, unzip it into your personal skills folder at ~/.claude/skills/. To make it available only inside one project, unzip it into that project's .claude/skills/ folder.
If you use the Claude apps instead of a local setup, you can upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: the .claude/skills/ folder inside your project
- Claude apps: Settings → Capabilities → Skills, then upload the skill
Trigger Rex with a natural request
You do not need a special command. Describe the idea in plain language and say what you want back. Rex is designed for cases where essential product requirements or success criteria are unclear and need a specification before other work starts.
If you are technical and already include most of the details in your request, Rex skips the questions and goes straight to the report.
- Ask for a requirements report on a feature you have only roughly described
- Ask for user stories with acceptance criteria before a developer starts
- Ask for edge cases and open questions to be surfaced before planning
- Ask for an amendment when scope changes mid-project, so only changed sections are updated
Example: from a vague idea to a report
Suppose you write: "We want a way for users to share their dashboards." Rex would challenge vague words, ask a small number of focused questions (for example, who the users are, whether sharing is public or limited to teammates, and whether it involves personal data), and then produce a report.
The report follows a fixed structure: a summary, a MoSCoW feature list with an Out of Scope section, user stories with Given/When/Then criteria, constraints, edge cases and risk flags, and open questions marked blocking or not.
What Rex needs and how it runs
Rex lists no required dependencies. It does not run scripts and does not use the network. The package contains only SKILL.md.
The skill's output is a report for a person or another agent to read. Keep private data and secrets out of your requests, because the skill asks that reports exclude them.
- Requires: nothing listed
- Runs scripts: no
- Uses network: no
- Files: SKILL.md
Safety notes and limits
Rex flags possible GDPR, HIPAA or accessibility concerns, but it is not a legal or regulatory authority. Its output may be incomplete or inaccurate, so have a qualified person review anything compliance-critical.
The skill itself notes that AI agents may hallucinate or give incorrect guidance, and that generated code and architecture should be verified before production.
- Do not treat the compliance flags as legal advice
- Review open questions before handing the report to developers
- Keep secrets and private data out of prompts and reports
Tips for better reports
Give Rex the context you already have: the user, the platform, existing systems and any constraints you have already decided. The skill uses this to skip questions you have answered and to avoid suggesting a tech stack you have not chosen.
Ask for amendments rather than full rewrites when requirements change, so you can see what moved.
- State the platform (web, mobile, desktop, API, CLI) if you know it
- Name any stack or database you have already locked in
- Answer the three questions per round as fully as you can
- Pass the report, not your raw chat history, to the next step
What people use it for
- A founder turning a one-line feature idea into a spec a developer can estimate
- A developer listing edge cases such as empty states and invalid input before coding a new feature
- A team using a multi-agent squad where a planning step reads a requirements document
- Updating an existing spec with a REX REPORT AMENDMENT after a scope change
Frequently asked questions
Is the Rex skill free?
The source lists the skill under the MIT licence. It is published in the agentic-awesome-skills repository on GitHub, and the source does not mention prices or paid tiers.
What do I need to install the Rex skill?
The skill lists no required dependencies. You need a host that supports agent skills, such as Claude, and either a .claude/skills folder or the Settings → Capabilities → Skills upload option.
Where do I put the Rex skill folder?
For personal use, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. In the Claude apps, upload it under Settings → Capabilities → Skills.
Does Rex write code?
No. The skill explicitly avoids code, schemas, architecture and implementation suggestions. Its output is a written requirements report.
Does Rex run scripts or access the internet?
No. The skill runs no scripts and uses no network. The package contains only SKILL.md.
How is Rex different from just asking an AI to write requirements?
Rex follows a fixed process: it asks at most three clarifying questions per round, uses MoSCoW priorities, writes Given/When/Then acceptance criteria and returns a versioned report format. Its output is designed so other people or agents can build from it without guessing.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Rex suits product managers, founders and developers who need a clear, testable spec before building, and who work with teams that pass requirements between steps. Skip it when requirements are already clear, when you want code or architecture, or when you need a legally verified compliance review, and pick a tool or a qualified specialist for those jobs.
Download the skill