Multi-Role Engineering Assistant — Skills Repository

A role-based AI assistant for software delivery teams. Describe your task in plain language and the correct engineering role — Developer, QA Engineer, DevOps, Solution Architect, Business Analyst, Security Engineer, or Data Engineer — is automatically applied with its full set of procedures, templates, and quality gates.

---

Install as a VS Code Extension — 3 steps

Step 1 — Clone and build

git clone <this-repo-url>
cd skill_chatbot
.\build.ps1

build.ps1 installs npm dependencies, packages the .vsix, and installs it into VS Code automatically. The extension appears immediately in Ctrl+Shift+X under INSTALLED.

Step 2 — Set your Anthropic API key

Ctrl+Shift+P  →  Skills Chatbot: Set Anthropic API Key

Paste your sk-ant-... key. It is stored in VS Code SecretStorage — never written to disk or any file.

Step 3 — Open the chat

Ctrl+Shift+P  →  Skills Chatbot: Open Chat

Type any engineering task. The correct role (Developer, QA, DevOps, Architect, BA, Security, Data Engineer) is detected automatically and a structured response is streamed back.

Share with teammates: send them the .vsix file that build.ps1 creates (skill-chatbot-1.0.0.vsix). They install it with code --install-extension skill-chatbot-1.0.0.vsix or via Extensions → Install from VSIX.

---

Available Roles

Role	Slash Command	What it produces
Developer	/developer	Code, unit tests, PR templates, commit messages
QA Engineer	/quality-analyst	Test cases, test plans, defect reports, TSR
DevOps Engineer	/devops	CI/CD pipelines, IaC, runbooks, monitoring config
Solution Architect	/solution-architect	HLD, ADRs, C4 diagrams, NFR register, API contracts
Business Analyst	/buisness-analyst	User stories, BRD, RTM, acceptance criteria
Security Engineer	/security-engineer	Threat models, OWASP review, CVSS findings, SAST/DAST
Data Engineer	/data-engineer	ETL/ELT pipelines, data models, dbt, data quality rules

---

Option 1 — VS Code Extension (recommended)

The cleanest way to share and use these skills. Install the .vsix once — the chat panel and all skill files are bundled inside.

What you get after installing the extension

Feature	How to use
Built-in chat panel	Ctrl+Shift+P → Skills Chatbot: Open Chat
Auto role detection	Just type your task — no slash commands needed
7 roles, streaming responses	Powered by Claude (Anthropic API key required)
Install skills into any project	Ctrl+Shift+P → Skills Chatbot: Install Skill Files into Workspace
Works with GitHub Copilot too	Skill files are copied into .github/ when you use the install command

Build the extension (.vsix)

# 1. Install dependencies
npm install

# 2. Install the VS Code extension packager
npm install -g @vscode/vsce

# 3. Package into a .vsix file
vsce package
# → creates skill-chatbot-1.0.0.vsix

Install the .vsix in VS Code

Method A — Drag and drop: Drag skill-chatbot-1.0.0.vsix into the VS Code Extensions panel.

Method B — Command palette: Ctrl+Shift+P → Extensions: Install from VSIX... → select the file.

Method C — Command line:

code --install-extension skill-chatbot-1.0.0.vsix

First-time setup

After installing the extension:

1. Ctrl+Shift+P → Skills Chatbot: Set Anthropic API Key → paste your sk-ant-... key
2. Ctrl+Shift+P → Skills Chatbot: Open Chat → start chatting

Your API key is stored in VS Code's SecretStorage — never written to disk.

Publish to VS Code Marketplace (optional — share with anyone)

# One-time: set your publisher ID in package.json "publisher" field
# Then:
vsce publish

After publishing, anyone installs it with:

ext install YOUR_PUBLISHER_ID.skill-chatbot

---

Option 2 — One-Command Install (no extension needed)

Add these skills to any project without cloning this repository.

Option A — npx (any platform, no install required)

# Install into current directory
npx skill-chatbot

# Install into a specific project
npx skill-chatbot /path/to/your-project

# Install + make Claude Code slash commands available in ALL projects
npx skill-chatbot --global

Option B — curl / PowerShell (once this repo is on GitHub)

# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/YOUR_ORG/skill_chatbot/main/install.sh | bash

# With global Claude commands
curl -fsSL https://raw.githubusercontent.com/YOUR_ORG/skill_chatbot/main/install.sh | bash -s -- --global

# Windows (PowerShell)
irm https://raw.githubusercontent.com/YOUR_ORG/skill_chatbot/main/install.ps1 | iex

# With global Claude commands
.\install.ps1 -Global

Option C — Clone and install locally

git clone <this-repo-url>
cd skill_chatbot

# Install into another project
./install.sh /path/to/your-project          # macOS / Linux
.\install.ps1 -Target C:\path\to\project    # Windows

# Install + global Claude commands
./install.sh --global
.\install.ps1 -Global

What gets installed:

Location	Purpose
.github/copilot-instructions.md	Auto-loaded by GitHub Copilot — role detection active immediately
.github/skills/*/SKILL.md	Attach with #file: in Copilot Chat for deep per-role context
.claude/commands/*.md	Slash commands (/developer, /quality-analyst, etc.) for Claude Code

Global flag: --global / -Global copies the Claude Code commands to ~/.claude/commands/ so /developer, /devops, etc. work in every project you open — not just the target directory.

---

Repository Structure

skill_chatbot/
├── .github/
│   ├── copilot-instructions.md        ← GitHub Copilot plugin (auto-loaded)
│   └── skills/
│       ├── skills/SKILL.md            ← GitHub Copilot orchestrator
│       ├── developer/SKILL.md
│       ├── quality-analyst/SKILL.md
│       ├── devops/SKILL.md
│       ├── solution-architect/SKILL.md
│       ├── business-analyst/SKILL.md
│       ├── security-engineer/SKILL.md
│       └── data-engineer/SKILL.md
└── .claude/
    └── commands/
        ├── skills.md                  ← Claude Code orchestrator (/skills)
        ├── developer.md
        ├── quality-analyst.md
        ├── devops.md
        ├── solution-architect.md
        ├── buisness-analyst.md
        ├── security-engineer.md
        └── data-engineer.md

---

PART A — Claude Code Setup & Usage

Prerequisites

• Claude Code CLI installed (npm install -g @anthropic-ai/claude-code or via the desktop app)
• Access to an Anthropic API key or Claude.ai Pro/Team subscription
• VS Code (optional but recommended for the Claude Code extension)

---

Step 1 — Get the Repository

Option A — Clone (recommended for teams)

git clone <repository-url>
cd skill_chatbot

Option B — Use globally (for all your projects)

Copy the commands folder to your global Claude config so the slash commands work in any project:

# macOS / Linux
cp -r .claude/commands/* ~/.claude/commands/

# Windows (PowerShell)
Copy-Item -Recurse .claude\commands\* "$env:USERPROFILE\.claude\commands\"

After copying globally, the slash commands work in every project you open — not just this repo.

---

Step 2 — Open Claude Code

In VS Code:

1. Install the Claude Code extension from the VS Code Marketplace (search "Claude Code").
2. Open this repository folder in VS Code.
3. Click the Claude Code icon in the activity bar, or press Ctrl+Shift+P → "Claude Code: Open Chat".

In the Terminal:

cd skill_chatbot
claude

In the Claude Code Desktop App:

1. Open the app.
2. Use "Open Folder" and select the skill_chatbot directory.

---

Step 3 — Use the Assistant (Claude Code)

Option A — Auto-route with /skills

Type your task in plain language. The assistant picks the right role automatically.

/skills Write a login API in Python with JWT authentication
/skills Write test cases for the password reset flow
/skills Build a GitHub Actions CI/CD pipeline for a Node.js service
/skills Write an ADR for choosing between PostgreSQL and MongoDB
/skills Write a user story for the customer checkout feature
/skills Review the payment endpoint for OWASP vulnerabilities
/skills Design an ETL pipeline for order data from the CRM

Option B — Invoke a role directly

Skip auto-routing and call the role slash command you need:

/developer Implement account lockout after 5 failed login attempts
/quality-analyst Draft a defect report — payment form accepts expired cards
/devops Write a deployment runbook for the v2.1.0 release
/solution-architect Produce a C4 context diagram for the billing service
/buisness-analyst Write a user story for the customer profile update feature
/security-engineer Run a STRIDE threat model on the authentication service
/data-engineer Design an incremental ELT pipeline for the events table

Option C — Cross-role output

Ask for output from more than one role in a single request:

/skills Write the user story AND the matching test cases for the cart checkout feature

---

Step 4 — Read the Output

Every response follows this structure:

### Role
[Which role was applied]

### Task Understanding
[How the assistant interpreted your request]

### Execution
[Steps taken, referencing the role's phased workflow]

### Output
[The artefact — code, test cases, ADR, user story, pipeline YAML, etc.]

### Assumptions (if any)
[Any assumptions made]

---

Claude Code Tips

Tip	Detail
Be specific	"Write test cases for the login endpoint — valid credentials, wrong password, locked account" gets better output than "test login"
Chain roles	Ask for BA output first, then QA — each builds on the previous
Reference tickets	Include Jira IDs (e.g. "for PROJ-389") — the assistant will include them in PR templates and commit messages
Use in any project	Copy commands to ~/.claude/commands/ to use slash commands in any repo

---

PART B — GitHub Copilot Setup & Usage

Prerequisites

• GitHub account with Copilot access (Individual, Team, or Enterprise licence)
• VS Code with the GitHub Copilot and GitHub Copilot Chat extensions installed
• Repository cloned or accessible on GitHub

---

Step 1 — Get the Repository

git clone <repository-url>
cd skill_chatbot

Open the folder in VS Code.

---

Step 2 — Verify Copilot is Active

1. Open VS Code.
2. Check the bottom status bar — you should see the Copilot icon (looks like a GitHub logo). If it shows "Copilot" or a checkmark, you are signed in and active.
3. If not signed in: Ctrl+Shift+P → "GitHub Copilot: Sign In".

The .github/copilot-instructions.md file in this repository is automatically read by Copilot the moment you open the repo. No manual step is needed — role detection and structured output are active immediately.

---

Step 3 — Open Copilot Chat

Platform	How to open
VS Code	Ctrl+Alt+I (Windows/Linux) or Cmd+Alt+I (Mac)
VS Code	Click the chat bubble icon in the activity bar
VS Code	Ctrl+Shift+P → "GitHub Copilot Chat: Focus on Chat View"
GitHub.com	Open the repository → click "Copilot" in the top navigation

---

Step 4 — Use the Assistant (GitHub Copilot)

Option A — Just type your task (auto-route)

The copilot-instructions.md is already active. Just describe your task:

Write a login API in Python with JWT authentication
Write test cases for the password reset flow
Build a GitHub Actions CI/CD pipeline for a Node.js service
Write an ADR for choosing between PostgreSQL and MongoDB
Write a user story for the customer checkout feature
Review the payment endpoint for OWASP vulnerabilities
Design an ETL pipeline for order data from the CRM

Copilot detects the role and applies it automatically.

---

Option B — Attach a specific skill file for precision

Type # in the chat input to open the file picker, then select the skill file:

#file:.github/skills/developer/SKILL.md
Implement account lockout after 5 failed login attempts

#file:.github/skills/quality-analyst/SKILL.md
Draft a test plan for the payment checkout flow

#file:.github/skills/devops/SKILL.md
Write a GitHub Actions CI/CD pipeline for a Node.js microservice

#file:.github/skills/solution-architect/SKILL.md
Write an ADR for choosing between PostgreSQL and MongoDB

#file:.github/skills/business-analyst/SKILL.md
Write a user story for the customer profile update feature

#file:.github/skills/security-engineer/SKILL.md
Run a STRIDE threat model on the authentication service

#file:.github/skills/data-engineer/SKILL.md
Design an incremental ELT pipeline for the events table

How to attach a file: In the Copilot Chat input box, type # → a file picker dropdown appears → start typing the filename (e.g. SKILL) → select the file → press Enter. The file is attached as context for your prompt.

---

Option C — Use @workspace for repo-grounded output

Combine @workspace with a skill file to ground the role output in your actual codebase:

@workspace #file:.github/skills/developer/SKILL.md
Find the orders service and add input validation to the create order endpoint

@workspace #file:.github/skills/security-engineer/SKILL.md
Review the authentication module for OWASP vulnerabilities

@workspace #file:.github/skills/data-engineer/SKILL.md
Find the existing ingestion jobs and add data quality checks to the orders pipeline

---

Option D — Cross-role output (attach multiple skill files)

Attach two or more skill files in one prompt for tasks that span roles:

#file:.github/skills/business-analyst/SKILL.md
#file:.github/skills/quality-analyst/SKILL.md
Write the user story AND acceptance criteria AND matching test cases for the cart checkout feature

#file:.github/skills/solution-architect/SKILL.md
#file:.github/skills/security-engineer/SKILL.md
Design the authentication architecture AND produce a threat model for it

---

Option E — Use on GitHub.com

1. Navigate to this repository on GitHub.com.
2. Click Copilot in the top navigation bar.
3. Type # to attach a skill file and then your request:

#.github/skills/devops/SKILL.md
Write a CI/CD pipeline definition for a containerised Node.js service

---

Step 5 — Copilot Agents Reference

Agent	What it does	When to use with a skill
@workspace	Searches your entire open repository	Grounding implementation or test tasks in real code
@github	Queries GitHub issues, PRs, code	Referencing open defects, tickets, or existing stories
@vscode	Answers VS Code settings questions	DevOps or tooling setup tasks
@terminal	Generates or explains terminal commands	Infrastructure and pipeline tasks

---

GitHub Copilot Tips

Tip	Detail
Auto-load is always on	copilot-instructions.md loads automatically — you do not need to attach anything for basic use
Attach for precision	Attaching the specific SKILL.md gives Copilot the full templates and quality gates for that role
Use @workspace for real code	Without @workspace, Copilot answers generically; with it, answers are grounded in your actual files
Multi-role in one prompt	Attach two skill files and ask for both artefacts — Copilot produces a labelled section per role
Re-attach on new chat	#file attachments apply to the current chat session; re-attach in a new chat window

---

PART C — Response Format (Both Platforms)

Every response — whether from Claude Code or GitHub Copilot — follows this structure:

### Role
[Developer / QA Engineer / DevOps Engineer / Solution Architect / Business Analyst / Security Engineer / Data Engineer]

### Task Understanding
[What the assistant understood you to be asking for]

### Execution
[The phased workflow steps followed to produce the output]

### Output
[The artefact: code, test cases, ADR, user story, pipeline YAML, threat model, data pipeline spec, etc.]

### Assumptions (if any)
[Any assumptions or constraints applied; "None" if not applicable]

---

PART D — Example Prompts by Role

Developer

Write a REST endpoint in Python (FastAPI) to create a new customer order with input validation
Implement retry logic with exponential backoff for the payment service HTTP client
Refactor the UserService class to follow SOLID principles
Write unit tests for the account lockout logic

QA Engineer

Write test cases for the login feature — valid login, wrong password, locked account, expired session
Draft a defect report: the checkout total is incorrect when a discount code is applied
Write a test plan for the Sprint 5 payment feature release
Define exit criteria for the regression cycle before go-live

DevOps Engineer

Write a GitHub Actions CI/CD pipeline for a Python FastAPI service using Docker
Write a Terraform module to provision an Azure App Service with a PostgreSQL database
Draft a deployment runbook for the v3.0.0 production release
Write a post-incident review for the 2026-04-20 payment service outage

Solution Architect

Write an ADR for choosing between PostgreSQL and MongoDB for a multi-tenant SaaS product
Produce a C4 context diagram for the order management system
Define NFRs for the payment service — performance, availability, and security
Design the integration between the CRM and the data warehouse

Business Analyst

Write a user story for the customer profile update feature with acceptance criteria
Draft BRD section 3 — functional requirements for the notification service
Assess the change request: the client wants to add real-time inventory updates to the checkout
Write a stakeholder status update for Sprint 5 requirements sign-off

Security Engineer

Review the login endpoint for OWASP Top 10 vulnerabilities
Write a STRIDE threat model for the payment processing service
Score this vulnerability with CVSS v3.1 and recommend remediation: unauthenticated access to admin API
Define secrets management for a Kubernetes-hosted microservices platform

Data Engineer

Design an incremental ELT pipeline to sync orders from PostgreSQL to Snowflake
Write a dbt model for the customer lifetime value report
Define data quality rules for the fact_orders table
Write a schema migration to add a soft-delete column to the customers table with rollback

---

PART E — Sharing with Your Team

Quickest path — send one command

Share this with your team and they are set up in under a minute:

# macOS / Linux — installs into current project AND globally
curl -fsSL https://raw.githubusercontent.com/YOUR_ORG/skill_chatbot/main/install.sh | bash -s -- --global

# Windows — installs into current project AND globally
irm https://raw.githubusercontent.com/YOUR_ORG/skill_chatbot/main/install.ps1 | iex; .\install.ps1 -Global

# Any platform via npx
npx skill-chatbot --global

For Claude Code (team-wide setup)

1. Share this repository URL with your team.
2. Each employee runs one of the install commands above — slash commands work immediately.
3. The --global flag means /developer, /quality-analyst, /devops, /solution-architect, /buisness-analyst, /security-engineer, and /data-engineer work in every Claude Code session, not just this repo.

---

For GitHub Copilot (zero setup for employees)

1. Share the repository with your team on GitHub (or your internal GitHub Enterprise instance).
2. Each employee clones the repo and opens it in VS Code.
3. That's it — copilot-instructions.md is automatically read by Copilot. No manual configuration.

For employees who work across many repositories and want role-routing everywhere:

1. In VS Code, open Settings (Ctrl+,).
2. Search for github.copilot.chat.codeGeneration.instructions.
3. Add a reference to the copilot-instructions.md file, or copy its content into a global Copilot instructions setting.

---

Troubleshooting

Problem	Solution
Slash command not found in Claude Code	Ensure you opened the skill_chatbot folder (not a parent folder). Commands only load from .claude/commands/ in the open project root.
Global slash commands not working	Verify the files were copied to ~/.claude/commands/ (macOS/Linux) or %USERPROFILE%\.claude\commands\ (Windows). Restart Claude Code after copying.
Copilot not applying role routing	Check that .github/copilot-instructions.md exists in the repo root. If you are on a branch, ensure the file is present on that branch.
Copilot giving generic answers	Attach the specific skill file using #file:.github/skills/<role>/SKILL.md for maximum precision.
@workspace not finding files	Ensure VS Code has indexed the workspace — open the Command Palette and run "GitHub Copilot: Refresh workspace index".
Cross-role output missing a section	Re-prompt with both skill files attached explicitly and state "produce output for both roles".

---

Adding New Roles

To add a new role:

1. Create .github/skills/<role-name>/SKILL.md — follow the template of any existing SKILL.md.
2. Create .claude/commands/<role-name>.md — follow the template of any existing command file.
3. Add the new role to the Intent Detection table in .claude/commands/skills.md.
4. Add the new role to the Skill File Registry and Intent Detection table in .github/skills/skills/SKILL.md.
5. Add the new role to the table in .github/copilot-instructions.md.
6. Update this README — add a row to the Available Roles table and a prompt example in Part D.