Product Management Skills - RFP to User Stories

Anthropic Agent Skills suite for converting RFP (Request for Proposal) or requirements documents into high-quality User Stories.

👤 Author

Po-chiang Chao

• 🌐 Connect with me on social media: bobchao.net
• 💼 Product Management mentoring: stableprogress.com

📁 Project Structure

(root)/
├── rfp-analyzer/                    # Skill 1: RFP Analysis
│   ├── SKILL.md                     # Main Skill definition
│   └── references/
│       ├── question-guidelines.md   # Question generation guidelines
│       └── invest-criteria.md       # INVEST principles explanation
│
├── story-writer/                    # Skill 2: Story Writing
│   ├── SKILL.md                     # Main Skill definition
│   ├── references/
│   │   └── story-templates.md       # Story template library
│   └── assets/
│       └── output-example.md        # Complete output example
│
├── story-refiner/                   # Skill 3: Story Refinement
│   ├── SKILL.md                     # Main Skill definition
│   ├── references/
│   │   └── evaluation-criteria.md   # Evaluation criteria
│   └── assets/
│       └── refine-example.md        # Refinement report example
│
└── README.md

🌐 Language Support

All Skills default to responding in the same language as user input. Users can explicitly specify a preferred language:

• "請用中文回答" → Responds in Chinese
• "Reply in Japanese" → Responds in Japanese
• No specification → Matches input document language

🎯 Design Philosophy

Division of Three Skills

Why This Design?

1. Separation of concerns: Analysis, writing, and refinement are different cognitive tasks
2. Automated correction: Refiner doesn't just report issues, it directly produces improved versions
3. Reduced manual intervention: Users only need final confirmation, not manual corrections
4. Flexible combination: Can choose workflow based on project complexity

Problems Solved

🚀 Usage Methods

Method 1: Quick Mode

Simplest usage method, directly produce Stories:

RFP → [story-writer] → User Stories Draft

Suitable for:

• Small to medium projects
• Quick results desired
• Will use Story Refiner separately if refinement is needed

Method 2: Complete Workflow

Suitable for complex projects or situations requiring manual review:

Step 1: RFP Analyzer
────────────────────
Input: Raw RFP document
Output: Structured analysis report
     ├── Stakeholder list
     ├── Functional module decomposition
     ├── System constraints
     └── Classified clarification questions

        ↓ [Manual review + answer 🔴 questions]

Step 2: Story Writer
────────────────────
Input: Analysis report + question answers
Output: User Stories draft

        ↓ [Optional: manual review]

Step 3: Story Refiner
────────────────────
Input: User Stories
Output: Refinement report
     ├── Quality assessment
     ├── Corrected Stories
     └── Final complete list

Suitable for:

• Large complex projects
• Phase-by-phase review needed
• Multi-person collaboration

Method 3: Direct Story Writing

Skip analysis, directly produce Stories:

RFP → [story-writer] → User Stories Draft

Suitable for:

• Simple, clear requirements
• Time pressure
• MVP/POC projects

Note: Can optionally use Story Refiner afterward for quality refinement.

📋 Detailed Skill Descriptions

1. RFP Analyzer

Trigger: Receive RFP or requirements document, need to "understand the full picture" first

Core Functions:

• Phase 1: Quick scan and background understanding
• Phase 2: Stakeholder identification
• Phase 3: Functional module decomposition
• Phase 4: Non-functional requirements extraction
• Phase 5: Question generation (three-tier classification)

Question Classification:

2. Story Writer

Trigger: Have analysis results or requirements list, need to convert to User Stories

Core Functions:

• Standard format writing (role/action/value)
• Granularity control
• Implied requirement handling (with boundaries)
• Acceptance criteria writing

Note: For quality refinement and automatic correction, use the separate Story Refiner skill after Story Writer completes its output.

3. Story Refiner

Trigger: Have Stories needing quality assurance and correction

Core Functions:

• Three-perspective evaluation (Developer/QA/Stakeholder)
• 1-5 point scoring system
• Auto-correct low-scoring Stories
• Iterative refinement (max 3 rounds, ensures split Stories are evaluated too)

Scoring Criteria:

Iterative Refinement:

Round 1: Evaluate all → Correct low-scoring items → May generate new Stories (splits)
    ↓
Round 2: Evaluate "corrected" + "newly generated" → Correct again if needed
    ↓
Round 3: (If still issues) Final fine-tuning
    ↓
Terminate: All Stories ≥ 4 points, or limit reached

Termination Conditions (stop when any is met):

1. All Stories score ≥ 4
2. No corrections this round
3. Already executed 3 rounds

📖 Reference File Descriptions

🔧 Customization

Adjust Question Quantity/Strictness

Modify rfp-analyzer/references/question-guidelines.md:

• Question quantity guidelines table
• Exclusion criteria list

Adjust Story Format

Modify story-writer/SKILL.md:

• Output format section
• Acceptance criteria format

Adjust Scoring Thresholds

Modify story-refiner/references/evaluation-criteria.md:

• Default threshold settings
• Strict/lenient mode definitions

Add Domain Templates

Add commonly used scenarios in story-writer/references/story-templates.md.

💡 Best Practices

DO ✅

• Use RFP Analyzer first to confirm understanding for complex projects
• Use refine mode to ensure Story quality
• Answer blocking questions before producing Stories
• Adjust template wording based on project

DON'T ❌

• Don't expect "zero manual intervention" (final confirmation still needed)
• Don't skip blocking questions and produce directly
• Don't mechanically apply all templates
• Don't ignore Refiner's correction suggestions

🔄 Version History

🔗 Related Resources

• Anthropic Skills Official Spec
• INVEST Principles Original Article
• User Story Best Practices

📄 License

This project is licensed under the MIT License.

MIT License

Copyright (c) 2026 Po-chiang Chao

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

For more information, visit: MIT License