DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to map an app's attack surface with parallel child agents with RiftX

Crawl an in-scope app into an inventory, then split review work across child agents on a shared board.

Authorization notice. RiftX is dual-use. Only use it on software, sites and systems you own or have explicit permission to test.

Who it is forA security tester starting an engagement on a large web application who wants a complete map before deep testing.

The problemMapping links, forms, API routes, and auth boundaries by hand is slow, and parallel review is hard to coordinate across separate chats.

The outcomeA structured attack-surface inventory and a board of review tasks, each claimed by a child agent and returned for main-agent review.

What you need first

  • Written authorization from the owner of the target system, with a defined scope
  • A local machine where you can run the RiftX app (the README badges reference Node.js, Next.js, and React)
  • A model provider key for OpenAI, Anthropic, Google, or a compatible endpoint
  • Playwright and Chromium available for the browser tool

Steps

  1. Open a session scoped to the authorized target and perform the first navigation in the browser.
  2. Run the crawl tool to collect links, forms including hidden fields, API routes from JS bundles, and authentication boundaries.
  3. Review the inventory, then ask the main Agent to propose review tasks on the shared task board.
  4. Approve the tasks so child Agents can claim them and run one work item at a time.
  5. Review each child result in the Agents and Messages views before marking the collaboration round finished.

Prompt for your agent

In RiftX, run the crawl tool on the authorized target <target-url> and build a board of review tasks from the inventory. Return the task list with owners, status, and evidence references. Verify that no task is marked complete until the main Agent has accepted the child result.

Caveat. The crawl only covers what the scoped browser can load, and the default limits of 32 admitted work items and 60 autonomous wakes apply to a single collaboration round.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for Ch1nfo/RiftX and "Map an app's attack surface with parallel child agents"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer