---
title: Tester Agent
description: The Tester agent role, test synthesis, and in-container test execution.
sidebar:
  order: 4
  label: Tester Agent
---

## The Tester Agent (`◉`)

The **Tester** verifies that the Coder's modifications actually work in practice.

Unlike tools that rely on hypothetical code analysis, NIKI's Tester generates test suites and **executes real commands inside the sandbox container**.

---

## What the Tester Does

1. **Test Generation:** Writes unit and integration test cases targeting the new functionality, edge cases, and boundary conditions.
2. **Containerized Execution:** Runs the project's real test runner (e.g. `npm test`, `cargo test`, `pytest`) inside the isolated container.
3. **Execution Analysis:** Captures stdout, stderr, exit codes, and duration.
4. **Artifact Production:** Formats the test run into a `TestReport` JSON artifact containing:
   * Total tests run, passed, and failed.
   * Captured failure traces.
   * Coverage assessment.

---

## Cost Optimization Tip

Because the Tester's primary job is generating standard test boilerplate and verifying test runner outputs, you can often use faster and more economical models here:

```toml
[agents.tester]
provider    = "openai"
model       = "gpt-4o-mini"
max_tokens  = 4096
temperature = 0.1
```