
# Continuous Pentesting

Continuous Pentesting automatically retests your application after changes.

It uses your last full assessment as the base configuration. It then runs smaller, targeted follow-up tests as your code evolves.

## When to use it

A full [Aikido Pentest](https://help.aikido.dev/pentests/aikido-pentest) is an on-purpose assessment of a running testing environment. Continuous Pentesting is the follow-up: after that first assessment, it retests the running app when you deploy or on a schedule.

Use it when you are in a heavily regulated industry such as finance or health tech, or when you have a complex setup with connected systems.

It is not a replacement for [Deep Review](https://help.aikido.dev/deep-review/how-deep-review-works). Deep Review reads the pull request in source (security and quality, 1 credit per PR) and does not send traffic at a live target. Continuous Pentesting does, and it is usage-based.

For the full map, see [Which Aikido product should I use?](https://help.aikido.dev/getting-started/which-aikido-product-should-i-use).

## Pricing and usage

Continuous Pentesting is usage-based. Each run is based on how many agents actually launch, and each agent costs 10 credits.

To preview your expected cost:

1. Open the [Pricing page](https://app.aikido.dev/ai-pentests/pricing?regionRedirect=1).
2. Select the repositories to include in your estimate.

The Pricing page shows an estimated yearly credit cost for Continuous Pentesting. **1 credit = $1 USD.**

This is an annualized projection based on your Git activity from the last 30 days. It can change as your activity changes.

On every configured deploy or schedule, Aikido first reviews the changes. It launches the agents needed for security-relevant changes and charges credits based on the final number launched. Larger changes can launch more agents, while smaller changes usually launch fewer. Documentation, test-only, and other non-security changes are skipped at no cost.

{% hint style="info" %}
**Rule of thumb:** run cost = `agents launched × 10 credits`.
{% endhint %}

You can cap that cost with a **Spending Limit**. Leave it empty for no per-run cap. See [Set a spending limit](#set-a-spending-limit).

See the [Pricing page](https://app.aikido.dev/ai-pentests/pricing?regionRedirect=1) for your annualized estimate and [Pentest Pricing](https://help.aikido.dev/pentests/pentest-pricing) for all pricing options.

The **Run now, pay later** no-pay condition (no High or Critical findings means you don't pay) applies only to your **first full assessment**. It does not apply to Continuous Pentesting runs. See [Run now, pay later](https://help.aikido.dev/pentests/pentest-pricing#run-now-pay-later) for details.

## Enable Continuous Pentesting

### Prerequisites

Before you enable Continuous Pentesting, make sure:

* You have the **Manage Pentests** permission.
* You completed at least one full assessment for this project.
* The full assessment has repositories connected.
* Your wallet balance meets the minimum required to launch agents.

### Configure the Frequency

1. Go to **Pentests** and open your project.
2. Click **Continuous Testing** in the project header.
3. Choose your frequency:
   * **Run on every deploy**
     * Select which branch should trigger testing for each connected repository.
     * Set a **Start Delay** to bundle multiple commits into one run.
   * **Run on schedule**:
     * Daily, weekly, monthly, or a custom interval.
     * The schedule starts when you enable Continuous Pentesting.
4. Optionally set a **Spending Limit**: the maximum credits one run can spend. Leave it empty for no per-run cap.
5. Click **Enable Continuous Testing** and confirm.

<div data-with-frame="true"><figure><img src="https://help.aikido.dev/_assets/4c97246d8022/Screenshot-2026-02-23-at-17.27.39.png" alt=""><figcaption></figcaption></figure></div>

## Set a spending limit

**Spending Limit** caps the credits a single continuous run can spend. Leave it empty for no per-run cap.

When a run starts, Aikido still reviews the code changes and decides how many agents those changes need:

* If that number is within your limit, every needed agent launches. You pay `agents launched × 10 credits`.
* If the changes need more agents than the limit allows, Aikido launches only up to the cap. It assigns those agents to the highest-priority changes first, so lower-priority parts of the change may not be tested on that run. You pay for the agents that launch, which will not exceed the limit.
* If the changes are not security-relevant, no agents launch and the run costs nothing. The limit does not create a minimum charge.

## How it works

Each cycle follows this flow:

1. Aikido checks whether your project is eligible to run.
2. It compares the latest commits on your configured branches with the last tested baseline.
3. If the changes should trigger a run, Aikido starts a new continuous assessment and analyzes the relevant code delta.

To avoid noisy or wasteful runs, Aikido applies guardrails:

* It waits for your configured **Start Delay** before triggering.
* It does not start a new run while another continuous run is still active.
* It requires the minimum wallet balance to launch agents.
* If you set a **Spending Limit**, it launches at most the number of agents that fit in that limit.

## Why a run might not start

If you expected a run but none started, check:

* No full assessment exists yet for this project.
* Wallet balance is below the minimum required to launch agents.
* No new commits were detected on your configured branches.
* The latest commit is still within your **Start Delay** window.
* Another continuous run is still in progress.
