Claude Code                         Opencode CLI + Kimi K2.5
    |                                   |
    +-- Write TASK.md ------------------+
    |   (test/review requirements)      |
    |                                   |
    +-- Execute: opencode run -m        |
    |   opencode/kimi-k2.5-free         |
    |   "Read TASK.md..."               |
    |                                   |
    |                                   +-- Reads TASK.md
    |                                   +-- Generates tests/review
    |                                   +-- Writes OUTPUT.md
    |                                   |
    +-- Read OUTPUT.md <----------------+
    |   (tests + findings + results)    |
    v                                   v

opencode run -m opencode/kimi-k2.5-free \
  "Read TASK.md in the current directory. Follow the testing instructions. Write all results to OUTPUT.md."

opencode run -m togetherai/moonshotai/Kimi-K2.5 \
  "Read TASK.md in the current directory. Follow the testing instructions. Write all results to OUTPUT.md."

# Start opencode TUI with Kimi K2.5 (Opencode)
cd /path/to/project && opencode . -m opencode/kimi-k2.5-free

# Or with Together.ai
cd /path/to/project && opencode . -m togetherai/moonshotai/Kimi-K2.5

# Create a prompt file
echo "Generate unit tests for the Calculator class" > prompt.txt
cat prompt.txt | opencode run -m opencode/kimi-k2.5-free

# Task: [Test Generation / Code Review / QA]

## Objective
[Clear statement of testing/review goal]

### Examples:
- Generate unit tests for user authentication module
- Review payment processing code for security issues
- Create integration tests for API endpoints
- Perform QA verification of search functionality

## Code to Test/Review

### File: src/auth/login.ts
```typescript
// Paste code here

// Paste supporting code

## Output File Template (OUTPUT.md)

Instruct Opencode with Kimi K2.5 to produce:

```markdown
# Results: [Task Title]

## Summary
[Brief overview of what was accomplished]

## Generated Tests (if applicable)

### Test Files Created
| File | Description | Coverage |
|------|-------------|----------|
| `tests/auth/login.test.ts` | Unit tests for login | 95% |
| `tests/auth/validate.test.ts` | Validation tests | 88% |

### Test Scenarios

#### Happy Path
1. [Test case description]
   - Input: [example input]
   - Expected: [expected output]

#### Edge Cases
1. [Edge case description]
   - Input: [boundary value]
   - Expected: [expected behavior]

#### Error Cases
1. [Error scenario]
   - Input: [invalid input]
   - Expected: [error handling]

## Review Findings (if applicable)

| Severity | Location | Issue | Recommendation |
|----------|----------|-------|----------------|
| CRITICAL | login.ts:45 | SQL injection | Use parameterized queries |
| HIGH | auth.ts:23 | Missing rate limiting | Add rate limiter |
| MEDIUM | users.ts:78 | No input validation | Add Zod validation |
| LOW | utils.ts:12 | Magic number | Extract constant |

## Code Coverage Analysis
- **Overall:** 87%
- **Critical paths:** 95%
- **Uncovered lines:** [list of uncovered code sections]

## Issues Encountered
- [Any problems and resolutions]

## Recommendations
1. [Priority action items]
2. [Secondary improvements]
3. [Testing best practices to adopt]

## Session
Session ID: `<session_id>` (for follow-up)

# Task: Generate Unit Tests for User Authentication

## Objective
Create comprehensive unit tests for the user authentication module covering login, registration, and password reset.

## Code to Test

### File: src/auth/auth.service.ts
```typescript
export class AuthService {
  async login(email: string, password: string): Promise<AuthResult> {
    const user = await this.userRepo.findByEmail(email);
    if (!user) throw new UnauthorizedError('Invalid credentials');
    
    const valid = await bcrypt.compare(password, user.passwordHash);
    if (!valid) throw new UnauthorizedError('Invalid credentials');
    
    const token = jwt.sign({ userId: user.id }, process.env.JWT_SECRET);
    return { token, user: this.sanitizeUser(user) };
  }
  
  async register(email: string, password: string): Promise<User> {
    const existing = await this.userRepo.findByEmail(email);
    if (existing) throw new ConflictError('Email already exists');
    
    const hash = await bcrypt.hash(password, 10);
    return this.userRepo.create({ email, passwordHash: hash });
  }
}

#### 2. Execute Opencode with Kimi K2.5

```bash
opencode run -m opencode/kimi-k2.5-free \
  "Read TASK.md in the current directory. Follow the instructions to generate comprehensive unit tests. Write all results to OUTPUT.md."

# Task: Security Review of Payment Processing

## Objective
Perform security-focused code review of payment processing module.

## Code to Review

### File: src/payments/payment.service.ts
[Code here...]

## Review Criteria
Rate findings: CRITICAL | HIGH | MEDIUM | LOW

### Security Focus
- Input validation
- SQL injection risks
- Authentication/authorization
- Sensitive data handling
- PCI compliance

## Context
- B2B SaaS application
- Handles credit card data
- SOC2 compliance required

## Output Format
Write security findings to OUTPUT.md with severity ratings and specific fix recommendations.

opencode run -m opencode/kimi-k2.5-free \
  "Read TASK.md, perform security review following the criteria, write findings to OUTPUT.md"

# For Opencode provider (Moonshot AI)
MOONSHOT_API_KEY=xxx      # Kimi models via Opencode (if needed)

# For Together.ai provider
TOGETHER_API_KEY=xxx      # Kimi models via Together.ai

# Alternative providers
OPENAI_API_KEY=xxx        # OpenAI models
GEMINI_API_KEY=xxx        # Google Gemini models