# heal Heal builds a production-like sandbox around an application, exercises its interfaces and services, and reports runtime defects with reproducible failing tests. It is designed to complement code review and can run locally or in the cloud. - Findings24: - Website: - Pricing: Freemium - Categories: AI Testing Assistant, Developer Tools ## What is heal? Heal is a software testing tool that runs an application in a production-like sandbox to find defects that may not be apparent from reviewing a pull request alone. It maps product features and entry points, sets up an environment with controlled data and substitutes for external services, then exercises the running product. Reported defects include reproduction steps and a test that demonstrates the failure, so teams can validate a fix. Heal is intended to complement code review by focusing on behavior, security, and performance issues encountered at runtime. ## What can heal do? ### Maps product features and entry points Heal scans a codebase to map features and how users or machines reach them, including interfaces such as a UI, API, CLI, SDK, webhooks, and scheduled jobs. ### Builds a production-like sandbox It learns how to run the application and sets up controlled data and runtime components, using mocks, stubs, or emulators where external dependencies cannot run directly. ### Exercises runtime behavior The agent drives application entry points and observes results such as screens, responses, exit codes, logs, stored data, and effects on third parties. ### Reports reproducible defects A reported issue includes reproduction steps and a test that fails against the running product; Heal re-runs the defect to check that it is reproducible. ### Validates security findings in context For security checks, Heal tests suspected vulnerabilities against the sandbox with an exploit and reports the observed result, rather than relying only on a code-pattern match. ### Supports local and cloud runs The product description says Heal can run locally or in the cloud, allowing teams to choose where to execute a hunt. ## Use Cases ### Check behavior beyond a pull request diff Run the product across features and shared data to look for defects involving unchanged code, configuration, or interactions that a review of changed lines may not exercise. ### Test external-service failure paths Use mocks or emulators to explore how an application behaves when payments, mail, or other third-party calls fail, slow down, or return duplicate responses. ### Investigate runtime security issues Exercise a suspected vulnerability in the sandbox and inspect what the attempted exploit actually returns, rather than treating a suspicious code pattern as a confirmed defect. ### Run broader checks before a release Teams can run Heal on the whole product periodically or before a release, then use its reproduction test to investigate reported issues and verify fixes. ## Frequently Asked Questions ### How is Heal different from a pull-request review bot? A review bot focuses on a diff and code changes. Heal runs the application in a sandbox and looks for behavior, security, and performance issues that occur at runtime. ### What does a reported defect include? Reports include reproduction steps and a test that demonstrates the failure. Heal says it re-runs a candidate and does not file what it cannot reproduce. ### How does Heal handle services that cannot run in the sandbox? It can mock, stub, or emulate dependencies, and replay scenarios such as failed, slow, or repeated service responses. ### Can Heal run locally? Yes. The product description says it runs locally or in the cloud. ## Quick Facts - **Product type**: Runtime application testing and defect detection - **Input**: An application codebase and its runtime environment - **Output**: Reproducible defect reports with failing tests - **Execution**: Local or cloud - **Website**: heal.dev