AI

Connect Your Paperclip Teams of AI Agents to the Web with Bright Data

Learn how Bright Data enables Paperclip AI agents to access web data, making your agent orchestration platform more powerful and capable.
11 min read
Paperclip Teams of AI Agents with Bright Data

In this tutorial, you will learn:

  • What Paperclip is and what it offers as an open-source AI orchestration platform for company management.
  • Why giving Paperclip agents access to the web can make the entire company more capable.
  • How Bright Data supports Paperclip through MCP and Agent Skills.
  • How to integrate Bright Data into your Paperclip company.

Let’s dive in!

What Is Paperclip?

Paperclip

Paperclip is an open-source platform for managing and orchestrating teams of AI agents. Its goal is to let you run agents as a coordinated company rather than as isolated tools. It is available on GitHub, where it boasts 78k+ stars.

In Paperclip, you can define business goals, create organizational structures, and assign roles and tasks. This work is then delegated across agents connected to local AI agent CLI harnesses, such as Claude Code, Codex CLI, and Gemini CLI.

Paperclip also provides scheduled heartbeats, goal alignment, budgets and cost controls, governance workflows, persistent task tracking, and detailed audit logs. Its use cases span software development, marketing, content creation, research, sales, support, QA, and more.

Why Your Paperclip Team of AI Agents Needs Web Access

Employees use the internet every day to do their jobs. They search for information, research competitors, check the latest news, find potential customers, compare products, and gather data before making decisions.

AI agents working in a company need to do the same. Without web access, your Paperclip agents are limited to the information available in their existing knowledge. This is where Bright Data comes in!

Bright Data provides AI-ready tools for live web search, source discovery, structured data extraction, browser automation, and dataset retrieval.

What makes Bright Data unique is its design for real-world workloads. Its global infrastructure includes more than 400 million residential IPs, providing the scale and anti-bot capabilities needed to access websites reliably. It supports unlimited concurrency, 99.99% uptime, and a 99.95% success rate.

How Bright Data Supports Paperclip: MCP + Agent Skills

Bright Data integrates with Paperclip via:

  • Bright Data MCP: An MCP server that gives AI agents access to 70+ tools for web search, scraping, data extraction, and browser automation.
  • Bright Data skills: A collection of Agent Skills that teach agents when and how to use Bright Data products.

Important: These two integrations are complementary and work best together. For instance, a specific skill helps Paperclip agents understand which Bright Data MCP tools to use for a given task and how to use them properly.

Bright Data MCP

The Bright Data MCP exposes 70+ tools. Even in Rapid mode (free tier with 5,000 requests per month), these include options for common web tasks:

Tool Description
search_engine + batch version Retrieves results from search engines such as Google, Bing, and Yandex.
scrape_as_markdown + batch version Converts web pages into clean Markdown while handling bot protection.
discover Performs AI-powered searches and returns ranked, relevant web results.

The remaining tools are more premium and include:

  • Web data feed tools to get structured data from Amazon, LinkedIn, Yahoo Finance, YouTube, Zillow, Instagram, Google Maps, and 40+ other domains.
  • Browser automation tools to interact with web pages.

Bright Data Agent Skills

The Bright Data agent skills give Paperclip agents specialized knowledge for using Bright Data products and web data workflows.

The available skills include:

Skill Description
agent-onboarding Helps agents configure Bright Data integrations and choose the appropriate setup.
brightdata-cli Guides agents in using the Bright Data CLI for web scraping and extraction.
data-feeds Extracts structured data from 40+ websites.
bright-data-mcp Explains how to use Bright Data MCP tools for web tasks.
bright-data-best-practices Provides guidance on Bright Data APIs and integrations.
competitive-intel Performs competitor analysis using live web data.
brand-listening Analyzes brand mentions and audience sentiment.
price-comparison Compares prices across online retailers.

Paperclip + Bright Data: Integration Approaches

Paperclip acts as the control plane for your AI company, organizing agents, goals, tasks, budgets, and governance. Bright Data can extend those agents with enterprise-grade web access tools.

Paperclip + Bright Data integration architecture

At the AI harness level, there are two main options:

  • MCP: Connect the agent to the Bright Data MCP server to gain access to 70+ web tools.
  • Agent Skills: Install Bright Data agent skills to give the agent the knowledge and instructions it needs to get the most from Bright Data’s products.

Paperclip will then manage these Bright Data-enhanced agents, without being aware of the integration.

At the Paperclip level, you can also install the Bright Data skills and assign them directly to some agents.

How to Integrate Bright Data into Paperclip

In this guide, you will see how to connect your Paperclip agent to Bright Data through both MCP and Agent Skills integrations.

Prerequisites

Before you start, make sure you have the following installed and configured.

For Paperclip:

  • A UNIX-based operating system (Linux, macOS, WSL 2).
  • Node.js 20+ installed locally.
  • pnpm installed locally.
  • One or more coding agents supported by Paperclip installed locally (e.g., Claude Code, OpenAI Codex CLI, Gemini CLI, Grok Build, Hermes, OpenCode, Pi, or Cursor CLI). In this guide, we will use Claude Code.

For the Bright Data MCP integration:

For the Bright Data agent skills integration:

While not strictly required, it is also helpful to have the following knowledge:

Step #1: Install and Set Up Paperclip

To start the Paperclip CLI installation wizard, run:

npx paperclipai onboard --yes

This single command handles the entire installation process:

  1. Downloads Paperclip.
  2. Creates a configuration directory at ~/.paperclip/.
  3. Initializes an embedded database.
  4. Starts the Paperclip server at http://localhost:3100.

Note: The --yes flag accepts all default options automatically. If you run the command without --yes, you can customize options such as the deployment mode, database, and storage.

In the terminal, you will see output similar to:

The CLI output of the npx installation command

Your default browser will open a new page at:

http://127.0.0.1:3100/onboarding

The Paperclip onboarding process will begin. This asks for some basic information about your company so Paperclip can manage your organization.

Start by entering your company name, such as “AwesomeAI Inc”:

Giving your company a name

Next, describe your company’s mission. For example:

Democratize AI by delivering powerful SaaS products that automate routine work and make cutting-edge technology accessible to everyone
Describing your company’s mission

Note: If you are not sure what to write, consider switching to the “Help me figure it out” section and answering the questions provided. Paperclip will use your answers to help define the mission.

Finally, specify your company’s team lead, who represents the company CEO. By default, Paperclip calls this role “Chief of Staff”:

Define your team lead

Click “Next.” Your company is now configured.

The final step is to connect your Chief of Staff to an AI agent CLI harness. If you are not familiar with this technology, read our article on AI harnesses.

Select the AI agent CLI harness you installed locally (in this example, Claude Code):

Selecting Claude Code as the adapter for your “Chief of staff” agent

Click “Give it a heartbeat” to verify that Paperclip can connect to the selected AI CLI harness. If everything works correctly, you will see:

Final step of the onboarding process

Next, you will be redirected to your company’s Paperclip dashboard URL. In this example, that is:

http://127.0.0.1:3100/AWE/dashboard
The Paperclip dashboard

The Chief of Staff will start hiring new agents (employees) based on your company’s mission. These agents will fill different roles and gradually build out your company’s organizational structure.

Well done! Paperclip is now installed and configured locally.

Step #2: Configure Bright Data MCP in Your Agent CLI Harness

Paperclip runs on top of an AI agent CLI harness installed locally, which means it does not support MCP configurations. If you want the agents orchestrated by Paperclip to access Bright Data Web MCP, configure it in the underlying AI agent CLI harness.

For more guidance, refer to these official guides:

For instance, when setting up the Bright Data MCP in Pro mode in Claude Code, you should see:

Note the Bright Data Web MCP connection in Claude Code

You can also configure the Bright Data skills directly at the AI CLI harness level. Still, remember that Paperclip supports agent skills natively. So, see how to configure the Bright Data skills in Paperclip in the next steps!

Step #3: Install the Bright Data Agent Skills in Paperclip

Start by selecting the “Skills” option from the left sidebar:

Reaching the “Skills” page

On the “Skills” page, click “New” and select “Import from path or URL”:

Selecting the “News > Import from path or URL” option

The “Import a skill” modal will open. Enter the following skills.sh URL:

https://www.skills.sh/brightdata/skills

That is the URL for Bright Data skills on skills.sh.

Importing the Bright Data skills from skills.sh

Click “Import” to add the skills to Paperclip. Most Bright Data skills rely on custom executable shell scripts. Paperclip does not currently support these scripts, so only a subset of the Bright Data skills will be imported:

  • agent-onboarding
  • brand-listening
  • brd-browser-debug
  • bright-data-best-practices
  • bright-data-mcp
  • brightdata-cli
  • competitive-intel
  • data-feeds

On the Paperclip “Skills” page, you should now see the aforementioned skills:

Note the Bright Data skills

Great! The Bright Data skills are now available in Paperclip.

Step #4: Assign Bright Data Skills to an Agent

The brightdata-cli skill teaches an agent how to use the Bright Data CLI. The CLI exposes commands for web scraping, search, discovery, and browser automation powered by Bright Data products.

Since you have already installed the Bright Data CLI, you can assign this skill directly to a Paperclip agent. Start by clicking the brightdata-cli skill card. On the skill page, click “Add to agent” and select “Chief of Staff”:

Adding the “brightdata-cli” skill to the “Chief of staff” agent

The CEO of the Paperclip company can now use the Bright Data CLI for web access, regardless of the tools available in the underlying AI CLI harness integration.

Fantastic! Time to see how your Chief of Staff can use Bright Data.

Step #5: Test the Integration

To verify that the Paperclip + Bright Data integration works, give the Chief of Staff a task that requires web search and scraping.

Click “New Task” and specify a task like this:

  • Title: “Identify AI C-Level Roles”
  • Description: “Use Bright Data CLI to search online for the most relevant AI-related C-level roles a SaaS company like AwesomeAI Inc. should have and hire for.”
  • For: “Chief of Staff”
Assigning a  new task to the “Chief of staff” agent

To proceed, press the “Create Task” button to launch the task.

The log shows that the Chief of Staff loaded the brightdata-cli skill and used it to complete the task:

Note the use of the Bright Data CLI

In detail, the agent:

  1. Called the Bright Data CLI search command (powered by SERP API) to perform relevant search queries on Google (e.g., “most important AI C-level executive roles for a saas company”).
  2. From the retrieved structured SERPs, selected the most interesting sources.
  3. Retrieved their content in Markdown by calling the Bright Data CLI scrape command (via Web Unlocker API)

From the task document, you can see that the agent identified six potential C-level roles from the discovered web pages:

The discovered AI-related C-level roles

After analyzing these roles, the agent decided that only two should be hired:

The selected AI-related C-level roles to hire for

Et voilà! This simple test shows how Paperclip agents can use Bright Data skills (and MCP tools) to complete web-related tasks.

Conclusion

In this blog post, you learned what Paperclip is and how it enables you to orchestrate AI agents as a company. You saw two ways to integrate Bright Data:

  1. Connect the underlying AI harnesses to Bright Data’s MCP and Agent Skills.
  2. Install Bright Data skills directly in Paperclip and assign them to specific agents.

Just as employees need access to the web to complete their work, AI agents in Paperclip require the same capabilities. Integrating Bright Data with Paperclip gives your agents reliable, scalable access to the public web.

Sign up for a free account today and explore our full range of AI-ready web services available in the Bright Data ecosystem.

No credit card required
Antonello Zanini

Technical Writer

5.5 years experience

Antonello Zanini is a technical writer, editor, and software engineer with 5M+ views. Expert in technical content strategy, web development, and project management.

Expertise
Web Development Web Scraping AI Integration