# AI Agent Bugfix
URL: /docs/cookbooks/ai-agent-bugfix

An agent that debugs and fixes a failing test on its own, then gets independently verified.



<CookbookMeta difficulty="Intermediate" time="~10 min" primitives="[&#x22;Alineo.start()&#x22;, &#x22;agent.prompt()&#x22;, &#x22;agent.bash()&#x22;]" />

An AI agent that debugs a failing test and fixes the bug itself — inside its own sandbox, using
nothing but bash and a model.

## Setup [#setup]

<Steps>
  <Step>
    Start OpenSandbox in Docker (one-time setup):

    ```bash
    bunx alineo-cli init
    ```
  </Step>

  <Step>
    Get a free API key from [build.nvidia.com](https://build.nvidia.com) and export it:

    ```bash
    export NVIDIA_API_KEY=...
    ```
  </Step>
</Steps>

## Run it [#run-it]

<CookbookPlayground
  cwd="cookbooks/ai-agent-bugfix"
  repoHref="https://github.com/DrejT/alineo/tree/main/cookbooks/ai-agent-bugfix"
  commands="[
  { command: &#x22;bun install&#x22; },
  { command: &#x22;bun start&#x22; },
]"
  output="`Planting an off-by-one bug in calc.py...
Running pytest — 1 failed (as expected)

Agent: reading calc.py... found it — range(n) should be range(n + 1). Patching...
Agent: re-running pytest inside the sandbox — 2 passed. Reporting done.

Independent verification (agent.sandbox.exec, not the agent's own claim):
pytest ................. 2 passed ✔`"
/>

## What it does [#what-it-does]

<Steps>
  <Step>
    Plants a deliberate off-by-one bug in `calc.py` and a test that catches it.
  </Step>

  <Step>
    Runs `pytest` via `agent.bash()` to show the failure.
  </Step>

  <Step>
    Prompts the agent — via `alineo`'s `Alineo.start()` + `agent.prompt()` — to find and fix the bug
    itself, streaming its reasoning and tool calls as they happen.
  </Step>

  <Step>
    **Re-runs `pytest` independently of the agent** (via `agent.sandbox.exec()`) to verify the fix,
    rather than trusting the agent's own claim that it passed.
  </Step>
</Steps>

`agents/bugfix-agent.json` configures a Pi agent on `python:3.11-slim` using the NVIDIA NIM API.
Swap `provider`/`model` for anything in `@alineo-labs/model-providers` to use a different key.

<Callout type="warn" title="Never trust the agent's self-report">
  Step 4 is the important part of this recipe: never trust an agent's self-report that a fix worked
  — re-run the check yourself against the sandbox it was working in.
</Callout>

<Callout title="Where to go next">
  See the [Agent SDK docs](/docs/agent) for the full `alineo` API (`prompt`, `bash`, `steer`,
  `fork`, model switching, and more), and
  [examples/pi-agent](https://github.com/DrejT/alineo/tree/main/examples/pi-agent) for a tour of
  every command it exposes.
</Callout>