7-Step Solution Wizard
A guide to the inputs, examples, and common mistakes for each step of the 7-Step wizard.
How to Use the 7-Step Solution Wizard
The ClickEye wizard consists of 10 screens in total, 2 of which are handled automatically by AI. The user directly enters or selects on 8 of these screens.
Overall Flow
Step 1. Enter company information
↓ (click Next)
[AI Automatic] Generate solution prototypes (~30s)
↓
Step 2. Select a prototype
↓ (click Next)
[AI Automatic] PM recommendation analysis (~10s)
↓
Step 3. Select a PM
↓
Step 4. Review PM configuration
↓
Step 5. Configure agents & skills
↓
Step 6. Select a platform
↓
Step 7. Enter environment variables
↓
Step 8. Final review & project creation
Step 1 — Company Information & Solution Requirements
In this step, the AI gathers your company context and requirements so it can design a solution.
Input Fields
| Field | Required | Limit | Description |
|---|---|---|---|
| Company Name | ✅ | 200 characters | Company or team name |
| Company Size | ✅ | Select | Startup (1–10) / Small business (11–50) / SMB (51–200) / Mid-size (201–1,000) / Enterprise (1,000+) |
| Industry | ✅ | Select | IT, Finance/Fintech, E-commerce, Healthcare, Education, Manufacturing, Logistics, Marketing, Gaming, Other |
| Tech Stack | ⬜ | Multi-select | Python, TypeScript, React, Next.js, FastAPI, Docker, etc. (multiple selections allowed) |
| Core Product/Service | ✅ | 500 characters | The product or service your company currently sells or operates |
| Business Type | ✅ | Select | B2B / B2C / B2B2C / Internal tool |
| Company Description | ⬜ | 1,000 characters | Additional context such as team composition, development maturity, and current problems |
| Solution Description | ✅ | Min. 50 / Max. 2,000 characters | The field the AI relies on most |
Writing Examples
How to write a good Solution Description:
✅ Good example:
We want to automate the customer inquiries we currently manage in Excel.
Our sales team of 8 handles over 100 inquiries a day. We'd like automatic replies to
repetitive questions, and a Slack notification for urgent cases.
Integration with our CRM would be even better.
❌ Bad example:
We need AI automation.
Common Mistakes
- Writing the solution description too briefly: If it is under the 50-character minimum, the Next button stays disabled. Describe your specific workflow and the outcomes you want.
- Leaving the tech stack unselected: You can proceed without selecting one, but choosing the technologies your team currently uses helps the AI design a more suitable solution.
- Skipping the company description: It is not a required field, but noting your team size, current technical level, and constraints produces a far more accurate prototype.
[AI Automatic] Solution Prototype Generation
When you click Next, the AI automatically generates 3 solution prototypes. This takes about 30 seconds. No user input is required on this screen.
Step 2 — Prototype Selection
Choose the one of the 3 AI-generated solution candidates that best matches your project direction.
How to Read a Card
Each prototype card displays the following information:
- Name: The solution's name
- Type: SaaS / REST API / Full-stack / Internal tool / MVP / Custom
- Description: An overview of the solution
Common Mistakes
- Choosing "the one that looks most expensive": A complex solution isn't always better. Choose the one that matches your team size and technical level.
- If you don't like any of them: Go back to the previous step, write your solution requirements in more detail, and regenerate.
[AI Automatic] PM Recommendation Analysis
After you select a prototype, the AI analyzes a list of PMs (Project Manager AIs) optimized for that solution. This takes about 10 seconds.
Step 3 — PM Selection
From the list of PMs recommended by the AI, choose the PM AI that will lead your project.
How to Read a PM Card
| Field | Description |
|---|---|
| Avatar & Name | PM profile |
| Domain | The PM's area of expertise |
| Match Rate (%) | Fit with the selected prototype |
| Recommendation Reason | The basis for the AI's analysis |
| Success Rate / Completed Projects / Average Rating | Performance metrics |
Common Mistakes
- Choosing based on match rate alone: Also check whether the domain matches your project's industry.
- Proceeding without a selection: Selecting a PM is required. If none is selected, the Next button is disabled.
Step 4 — PM Configuration Review
In this step, you review which sub-agents and skills make up your selected PM. You can simply review and move on without making any changes.
Step 5 — Agent & Skill Configuration
Default agents are pre-selected based on the PM configuration. You can add or remove them as needed.
Agents
Each AI agent handles a specific role (writing code, testing, documentation, etc.). We recommend keeping the default selections.
Integration Skills (Optional)
Skills are integration features for connecting to external services.
| Skill | Description | Additional Setup Required |
|---|---|---|
| linear | Linear issue tracking integration | ✅ LINEAR_API_KEY, LINEAR_TEAM_ID |
| slack | Sending Slack notifications | ✅ SLACK_WEBHOOK_URL |
| github | Automatic GitHub PR creation | ✅ GITHUB_TOKEN |
When selecting the Linear skill:
LINEAR_API_KEYandLINEAR_TEAM_IDare added as required fields in Step 7 (Environment Variables). See the Linear Integration Setup Guide.
Common Mistakes
- Adding skills you don't need: Adding a skill also makes its API key a required input. Don't add skills for services you won't use.
Step 6 — Platform Selection
Select the AI agent platform that will run your downloaded ZIP.
| Platform | Recommended For | Notes |
|---|---|---|
| Claude Code ⭐ | Most users | Anthropic's AI coding agent; best supported |
| Gemini CLI | Users in the Google ecosystem | Based on Google Gemini |
| Cursor | Those who prefer an AI code editor | A VS Code-based editor |
| Codex | OpenAI users | Based on OpenAI Codex |
Common Mistakes
- Selecting Claude Code without an ANTHROPIC_API_KEY: If you select Claude Code, you must enter your Anthropic API key in Step 7. Get one in advance at console.anthropic.com.
Step 7 — Environment Variable Configuration
Enter the API keys and environment variables needed to run your solution. The values entered here are saved to the .env file inside the ZIP.
Required API Keys
| Key | Required | Where to Get It |
|---|---|---|
ANTHROPIC_API_KEY | Always required | console.anthropic.com |
LINEAR_API_KEY | When the Linear skill is selected | linear.app/settings/api |
LINEAR_TEAM_ID | When the Linear skill is selected | Found in your Linear team URL |
When the Linear Skill Is Selected — Real-Time Tracking Methods
| Method | Cost | Characteristics | Recommended For |
|---|---|---|---|
| Cloudflare Tunnel ⭐ | Free | Static URL, requires the terminal to stay open | Most users |
| ngrok | Paid $8/mo (fixed) / Free (temporary) | URL changes on restart (free plan) | Users with an ngrok account |
| 30-second polling | Free | No webhook needed, up to 30-second delay | When setting up a tunnel is difficult |
How to Add Environment Variables
- In the Add Environment Variable section at the bottom of the screen, enter the key name (
KEY_NAME) and value. - Click the + button or press Enter.
- If you want to enter a value later, extract the ZIP and edit the
.envfile directly.
Common Mistakes
- Proceeding without entering
ANTHROPIC_API_KEY: You can move to the next step even when a required key is unset, but an error will occur when you run the ZIP. Always enter it before proceeding. - Sharing the ZIP file: The
.envfile contains your actual keys. Do not share the ZIP file with others. - Choosing the Cloudflare method on the ngrok free plan: The URL changes on restart, invalidating the webhook. You would have to re-save the URL on the ClickEye settings page.
Step 8 — Final Review & Project Creation
This shows a summary of all the settings you have entered so far. If everything looks correct, click the Create Project button.
Items to Review
- Company name / Business type / Core product
- Selected prototype name & type
- Selected PM
- Agent & skill list
- Platform
- Number of configured environment variables
After Project Creation
- A ZIP file download button is displayed.
- Extract the ZIP and launch your selected platform (Claude Code, etc.).
- AI development begins.
Frequently Asked Questions
Q. If I go back to a previous step, does the AI regenerate? A. The prototype is regenerated only when you change the company information (Step 1). The other steps do not regenerate even if you go back.
Q. What happens if I close the wizard partway through? A. Your inputs are preserved as long as the browser session is maintained. If you close the browser completely, you'll have to start over from the beginning.
Q. What files are in the generated ZIP?
A. It includes the configuration files for your selected platform (.claude/, .gemini/, etc.), .env, CLAUDE.md, README.md, and more.