hackerai - Find and fix vulnerabilities by chatting with AI

项目说明

<p align="center">
<a href="
<img src="public/icon-512x512.png" width="150" alt="HackerAI Logo">
</a>
</p>

<h1 align="center">HackerAI</h1>

<h2 align="center">Your AI-Powered Penetration Testing Assistant</h2>

<di="center">

![Website]()

</div>

Getting started

Coding agents should start with AGENTS.md.

Prerequisites

You'll need the following accounts:

Required:

Optional:

Clone the repo

Na to the project directory

Install dependencies

Run the setup script

To use abliteration.ai for eligible security requests, create an API key in the
abliteration.ai console and set
ABLITERATION_API_KEY in .en, , and Trigger.de this
optional key, HackerAI continues using its standard models.

Start the de server

This runs both Next.js and Convex de:

Or run them separately in two terminals:

Run the Trigger.de

Agent mode runs the agent loop on a Trigger.dev task.
To use the agent locally:

1. Create a project at and copy your dev secret
key (tr_dev_…) into .en as TRIGGER_SECRET_KEY.
2. In the Trigger.de → your project → En ,
add the env vars the task needs to run (these live on the worker, not on
): NEXT_PUBLIC_CON, CON,
OPENROUTER_API_KEY, OPENAI_API_KEY, AWS_S3_ACCESS_KEY_ID,
AWS_S3_SECRET_ACCESS_KEY, AWS_S3_REGION, AWS_S3_BUCKET_NAME, and
E2B_API_KEY. Cloud Agent execution currently uses E2B. MIOSA execution and
migration are paused in code, including explicit provider o. Retain
MIOSA_API_KEY when cleanup or recovery of existing MIOSA files is needed.
Before resuming the rollout, follow the MIOSA runbook.
New Miosa
workspaces default to the native hackerai-tools template; optionally set
MIOSA_TEMPLATE_ID to o it. An existing miosa-sandbox-docker
o still selects the Docker template, so remove or update that value
in each intended runtime to use the nati. Existing workspaces
retain their original runtime and files. Migrated workspaces require
recovery before they can use E2B; their retained E2B source may be stale.
Add any optional keys you use
(ABLITERATION_API_KEY, PERPLEXITY_API_KEY, JINA_API_KEY, etc.).
3. Start the worker in a third terminal:

This starts the default Trigger.de used by local Agent requests.
To start an explicitly routed Trigger.de instead, set a stable
branch name with
TRIGGER_DE=my-local-agent pnpm de. Only use that o
when the request path is configured to target the same Trigger.de.

项目信息

开发语言：TypeScript

Star 数：739

技术要点

该项目属于漏洞知识库方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕Find and fix vulnerabilities by chatting with AI展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。