# Requirements Analysis

> Use when gathering and refining requirements into a structured TASK.

- Skill: `matrixfounder/requirements-analysis` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add matrixfounder/requirements-analysis`
- Raw SKILL.md: https://api.skillmd.com/api/skills/matrixfounder/requirements-analysis/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: MatrixFounder (https://skillmd.com/u/matrixfounder)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/matrixfounder/requirements-analysis

---

# Requirements Analysis

## 1. Analysis Loop
1. **Reconnaissance:** Read project structure, `.AGENTS.md`.
2. **Identification:** What are the use cases? Actors? Preconditions?
3. **Clarification:** List open questions. Better to ask now than fail later.

## 2. Technical Specification (TASK) Structure

You MUST follow the structure defined in `assets/task_template.md`.
Read this file to understand the required headers and sections.

## 3. Important Rules

### ✅ DO:
1. **Be detailed:** Describe every step in scenarios.
2. **Think about edge cases:** Consider errors, exceptions, boundary conditions.
3. **Ask questions:** If something is unclear — add to "Open Questions".
4. **Use existing terminology:** Use terms from project documentation.
5. **Link to existing functionality:** Explicitly state how new functionality interacts with existing.

### ❌ DO NOT:
1. **DO NOT write code:** You create TASK, not implementation.
2. **DO NOT design architecture:** This is the Architect's task.
3. **DO NOT invent:** If unclear, ask a question.
4. **DO NOT make assumptions:** Explicitly state where you make an assumption.
5. **DO NOT overcomplicate:** Keep it simple, don't overengineer.
6. **DO NOT ignore existing functionality:** Study the project before writing TASK.
7. **DO NOT leave important decisions for later:** All key decisions must be selected or clarification requested from user.

### 🔴 CRITICAL: Uncertainty Management
You are at the earliest stage of development. Unresolved uncertainty now can lead to project failure. Therefore:
1. **Pay maximum attention to unclear points**
2. **Do not hesitate to ask many questions**
3. **Better to ask a "stupid" question than make an incorrect assumption**
4. **If in doubt — add to "Open Questions"**