requirements-elicitation
Skill0 starsChanged 4 months agoMITView on GitHub
Guides the user through an interactive, conversational flow to elicit functional requirements (RF) and non-functional requirements (RNF) for a new feature or module. Outputs a complete, ready-to-use Feature Map in LBaC format saved to docs/features/. Trigger this skill when the user says things like: "I want to build X", "help me define requirements for Y", "new feature", "create a feature map", "what do I need for Z", or "let's define a new module".
5 other projects use a version of this.
Adopted by
Most recently taken up by
- mtarcure/claude-vibe-squadtheir own version, first seen Oct 28, 2026
- chinthakat/agent-skillstheir own version, first seen Sep 21, 2026
- jacarty/claude-toolkittheir own version, first seen May 23, 2026
- 45ck/business-analysis-skillstheir own version, first seen Mar 28, 2026
- Exploration-labs/Nates-Substack-Skillstheir own version, first seen Oct 18, 2025
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- mtarcure/claude-vibe-squad★ 220same name
- Exploration-labs/Nates-Substack-Skills★ 55same name
- 45ck/business-analysis-skills★ 28same name
- chinthakat/agent-skills★ 0same name
- jacarty/claude-toolkit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- .agents/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/lbac_rules.mdcCursor rule
The file
Show its rules and text
We split this skill into the rules it gives an agent
17 rules. 0 are common, 9 are distinctive, and 8 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
9
Distinctive
This author's own choices. Worth a look.
8
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
File exists →: Read it. Show the user the current state summary and ask:
- Distinctive: line 33
File does not exist →: Start a new session from Level 1, Phase 0.
- Distinctive: line 43
One phase at a time: Never ask multiple phases together.
- Distinctive: line 44
Summarize then advance: After each phase, confirm what you captured before moving on.
- Distinctive: line 45
Natural language mirroring: Use the user's own vocabulary.
- Distinctive: line 47
Pause points at level boundaries: After completing each level, ask the user if they want to continue to the next level or save and stop.
- Very specific: line 49
Save state incrementally: Update tmtp_session.md after every completed phase.
- Distinctive: line 50
Generate documents immediately: Don't wait until the end — generate each Feature Map as soon as its module is fully completed in Level 2.
- Distinctive: line 60
Create this file at the end of Level 1.
- Very specific: line 116
Update state: Phase 0 → [x]
- Very specific: line 160
If user says stop → update Resume From in state file and end session.
- Very specific: line 223
After Phase 5, generate docs/features/<module_name_snake_case>.md:
- Very specific: line 263
Update state: module → [x], add generated file to "Generated Documents".
- Distinctive: line 274
After every module, give the user a chance to stop:
- Very specific: line 280
When all modules are [x]:
- Very specific: line 313
After this phase, go back and add the NFR table to all Feature Maps generated in Level 2.
- Very specific: line 385
Update docs/LBAC.md — Add all Feature Maps, sprint plan, sprint history to index tables
“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill: Tell Me the Project — PM Kickoff
Goal
Act as a senior Project Manager conducting a progressive, multi-session project kickoff. Elicit everything a development team needs — vision, modules, RF, RNF, business rules, user stories, and roadmap — organized in three levels that go from general to specific.
A session can be completed in full or paused at any level boundary and resumed later using
a persistent state file: docs/planning/tmtp_session.md.
MANDATORY STARTUP — Check Session State
Before doing anything else, check if docs/planning/tmtp_session.md exists:
- File exists → Read it. Show the user the current state summary and ask:
"I found an existing session for [project_name]. Last updated: [date]. Resume from: [next pending item]? Reply yes to continue, or restart to begin a new session."
- File does not exist → Start a new session from Level 1, Phase 0.
Greet the user:
"Let's define your project together. I'll guide you through 3 levels — from the big picture down to engineering details. You can stop at any level boundary and resume later. Let's begin with the basics."
Core Behavior Rules
- One phase at a time. Never ask multiple phases together.
- Summarize then advance. After each phase, confirm what you captured before moving on.
- Natural language mirroring. Use the user's own vocabulary.
- Infer the obvious. "Users must log in" → capture JWT/auth NFR silently.
- Pause points at level boundaries. After completing each level, ask the user if they want to continue to the next level or save and stop.
- Save state incrementally. Update
tmtp_session.mdafter every completed phase. - Generate documents immediately. Don't wait until the end — generate each Feature Map as soon as its module is fully completed in Level 2.
- Never auto-generate ADRs. List triggers only; user fills them using Template 2.
Session State File
Path: docs/planning/tmtp_session.md
Create this file at the end of Level 1. Update it after every completed phase and module.
# Tell Me the Project — Session State
**Project:** [project_name]
**Last Session:** [YYYY-MM-DD]
**Resume From:** [next pending item description]
---
## Level 1 — Big Picture
- [ ] Phase 0: Project Identity
- [ ] Phase 1: Stakeholders & Roles
- [ ] Phase 2: Module Discovery
## Level 2 — Module Blueprints
*(populated after Level 1 is complete)*
- [ ] [module_name] → docs/features/[module_name].md
## Level 3 — Engineering & Planning
- [ ] Phase 6: Cross-cutting NFRs
- [ ] Phase 7: Roadmap & Prioritization
- [ ] Phase 8: User Stories & Backlog
- [ ] Phase 9: Architecture Triggers (Pending ADRs)
- [ ] Phase 10: Final document generation
## Generated Documents
*(updated as each document is created)*
- [ ] docs/planning/proposal_planning_qa.md
- [ ] docs/planning/sprint_history.md
- [ ] docs/business/business_rules.mdStatus symbols:
[ ]= Pending[/]= In Progress (started but not finished)[x]= Completed
─────────────────────────────────────────
LEVEL 1 — BIG PICTURE
Goal: Establish the project's north star and map its surface area.
─────────────────────────────────────────
Phase 0 — Project Identity
Ask:
"Let's start with the basics:
- What is the project called?
- What problem does it solve? (one sentence for a non-technical stakeholder)
- Who is it built for? (target user or customer segment)
- What does success look like in 6 months?"
Capture: project_name, elevator_pitch, target_customer, success_criteria
Update state: Phase 0 → [x]
Phase 1 — Stakeholders & User Roles
Ask:
"Who interacts with this product?
- List every user role (admin, end user, external API, guest...).
- For each: what is their main goal inside the system?
- Any roles with elevated permissions or restricted access?"
Build a table: | Role | Main Goal | Permissions |
Update state: Phase 1 → [x]
Phase 2 — Module Discovery
Ask:
"Let's map the product's main areas at a high level. What are the 3 to 7 main modules or feature areas this product needs? For each, give me a one-liner — what it does, not how it works.
Examples: User Authentication, Billing & Subscriptions, Notifications, Admin Dashboard."
List all modules clearly. Populate Level 2 section of state file with one entry per module.
Update state: Phase 2 → [x]
Level 1 — Pause Point
Show:
"✅ Level 1 complete. Session saved to
docs/planning/tmtp_session.md.Modules to define in Level 2: [list all modules]
Ready to dive into the module details? Reply yes to continue to Level 2, or stop to save and resume in a future session."
If user says stop → update Resume From in state file and end session.
─────────────────────────────────────────
LEVEL 2 — MODULE BLUEPRINTS
Goal: Define RF for each module, one at a time.
─────────────────────────────────────────
Process modules in the order listed in Level 2 of the state file. For each module, run Phases 3–5. Then generate the Feature Map before moving to the next module.
Phase 3 — Happy Path (per module)
Mark current module as [/] in state file.
For the current module, ask:
"Let's define [Module Name] in detail. Walk me through the perfect, successful scenario step by step:
- What does the user do first?
- What does the system respond with?
- How does it end successfully?"
Convert each step into a BDD scenario:
Given [precondition]
When [user action]
Then [system response]
And [additional outcomes]Capture: module.happy_path_bdd[]
Phase 4 — Error & Edge Cases (per module)
Ask:
"Now the failure modes for [Module Name]:
- What happens with invalid or missing input?
- What if a required external service fails (auth, payment, email, API)?
- Are there quotas, plan limits, or business rules that should block this?"
Generate BDD error scenarios for each case.
Capture: module.error_bdd[]
Phase 5 — Module Business Rules (per module)
Ask:
"Any specific rules or policies for [Module Name]?
- Usage limits per plan or role?
- Data ownership or visibility constraints?
- Pricing or billing rules tied to this module? If none, just say 'N/A'."
Capture: module.business_rules[]
Module Output — Generate Feature Map
After Phase 5, generate docs/features/<module_name_snake_case>.md:
# 🏷️ Feature: [Module Name]
[module description]
---
## 🏛️ "As Is" (Current State)
### 1. Behavior & UX
[greenfield or existing behavior]
### 2. Files & Infrastructure
[existing files or "Greenfield — no existing files"]
---
## 🚀 "To Be" (Target State)
### 1. New Requirements & UX
[prose of happy path + error flows]
### 2. Non-Functional Requirements (NFRs)
*(Cross-cutting NFRs to be added after Level 3 — Phase 6)*
### 3. Code & Structure Impact
- New Files: [TBD — to be detailed in Implementation Plan]
- DB Changes: [TBD]
---
## ✅ Acceptance Criteria (BDD)
[all BDD scenarios from Phases 3-4]
---
## 🛠️ Related Technical Debt
N/AUpdate state: module → [x], add generated file to "Generated Documents".
Notify user:
"✅ [Module Name] complete. Feature Map saved at
docs/features/[name].md. [N] modules remaining in Level 2: [list pending modules] Continue with [next module]?"
Level 2 — Pause Point (after each module)
After every module, give the user a chance to stop:
"Module [N/total] complete. Reply next to continue with [next module name], or stop to save and resume in a future session."
Level 2 — Completion Message
When all modules are [x]:
"✅ Level 2 complete. All [N] module Feature Maps have been generated.
Ready to define the engineering details and build the project roadmap? Reply yes to continue to Level 3, or stop to save and resume later."
─────────────────────────────────────────
LEVEL 3 — ENGINEERING & PLANNING
Goal: Cross-cutting NFRs, roadmap, backlog, architecture decisions.
─────────────────────────────────────────
Phase 6 — Cross-Cutting NFRs & Quality KPIs
Go through each category ONE BY ONE:
| Category | Prompt |
| Performance | "Acceptable response time for critical operations?" |
| Security | "Auth strategy? (JWT, OAuth2, SSO) Role-based access?" |
| Scalability | "Peak load? (concurrent users, tx/sec, data volume)" |
| Availability | "Uptime SLA? (99.9%, 99.99%, business hours only?)" |
| Data Retention | "How long stored? Archiving or deletion policy?" |
| Compliance | "Regulated data? (GDPR, HIPAA, PCI-DSS, SOC2)" |
| Accessibility | "UI needs to meet WCAG 2.1 AA?" |
| Observability | "Logging/tracing/alerting? (Sentry, Datadog, CloudWatch?)" |
| Internationalization | "Multi-language or multi-timezone support?" |
| Disaster Recovery | "RTO/RPO targets? Backup strategy?" |
| Code & Software Quality KPIs | "Target test coverage (≥80%)? Technical debt limits (<5%)? Security rules (0 high/critical)? BDD pass rate (≥85%)? PM KPI agreement requirements?" |
Build NFR table: | Category | Requirement | Target / SLA |
After this phase, go back and add the NFR table to all Feature Maps generated in Level 2. Also prepare the Quality & Code Quality KPIs Dashboard for docs/planning/proposal_planning_qa.md.
Update state: Phase 6 → [x]
Phase 7 — Roadmap & Prioritization
Ask:
"Let's prioritize:
- Which modules are required for the MVP?
- Which are Phase B (growth) or Phase C (scale)?
- Any hard deadlines or external dependencies?
- How long is each sprint for your team? (1 or 2 weeks?)"
Generate phased roadmap:
Phase A — MVP: [modules] — Sprint 1–N
Phase B — Growth: [modules] — Sprint N+1–M
Phase C — Scale: [modules] — TBDUpdate state: Phase 7 → [x]
Phase 8 — User Stories & Backlog
For each module (MVP first), generate numbered User Stories:
[US-001]: [Title]
As [role], I want [capability], so that [value].
Sprint: [N] | Phase: [A/B/C] | Estimate: [pts]
BDD: → [Scenario names]Show backlog table: | US-ID | Title | Role | Phase | Sprint | Points |
Update state: Phase 8 → [x]
Phase 9 — Architecture Triggers (Pending ADRs)
Review full project scope. Detect decisions that need an ADR:
| Trigger | Example question |
| Database | Relational vs. document? Multi-tenant? |
| Auth | JWT vs. session vs. OAuth2 provider? |
| Communication | Sync REST vs. async events/queues? |
| Deployment | Monolith vs. microservices vs. serverless? |
| Integrations | Vendor lock-in tradeoffs? |
| Caching | Redis? In-memory? CDN? |
| File storage | S3, GCS, local filesystem? |
Output for each trigger:
⚠️ Pending ADR: [Decision Title]
Why it matters: [one-line rationale]
Template: docs/templates/adr_template.mdUpdate state: Phase 9 → [x]
Phase 10 — Final Document Generation
Generate:
docs/planning/proposal_planning_qa.md— Roadmap, sprint tables, QA strategy, Quality & Code Quality KPIs dashboard, PM alignment guide, DoDdocs/planning/sprint_history.md— Sprint 0 kickoff + Sprint 1 planned storiesdocs/business/business_rules.md— Pricing tiers + BR register (if applicable)- Update
docs/LBAC.md— Add all Feature Maps, sprint plan, sprint history to index tables
Mark all items [x] in state file. Set Resume From: COMPLETE.
Level 3 — Closing Message
"🎉 Tell Me the Project — Complete.
Generated:
- [N] Feature Maps in
docs/features/- Sprint plan →
docs/planning/proposal_planning_qa.md- Sprint history →
docs/planning/sprint_history.md- [Business rules →
docs/business/business_rules.md]docs/LBAC.mdindex updatedPending ADRs — create manually using Template 2: [list each pending ADR]
Next step: Review the Feature Maps, then use Template 4 (Implementation Plan) to begin coding the first MVP user stories."
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

