FeaturedLegal & Compliance#MCP#Web Research#Fetch#Agent#Prompt Injection

Governed Web Research Agent (MCP)

Let an Agent fetch and summarize web pages through a Fetch MCP server routed behind the Arc Gate MCP governance proxy.

Workflow at a glance

The full canvas, before you import it

Click any node to see its config.

#MCP#Web Research#Fetch#Agent#Prompt Injection

Click a node to select it — same as the Heym editor; the panel shows its settings.

3 nodes · Free & source-available

Governed Web Research Agent (MCP)

This template shows an Agent that fetches and summarizes web pages, with every fetched page routed through a governance proxy before it reaches the model. The Agent's Fetch MCP server (uvx mcp-server-fetch) runs behind Arc Gate MCP (uvx arc-gate-mcp), a runtime proxy that screens tool results for prompt-injection and capability abuse.

What this workflow does

  1. ResearchQuestion provides a URL or research prompt
  2. WebResearchAgent calls the Fetch MCP tool; the page passes through Arc Gate's browser_agent policy first
  3. The Agent summarizes the governed content and notes anything Arc Gate flags or blocks
  4. ResearchSummary returns the final summary

Use cases

  • Summarizing untrusted web pages where fetched content could hide instructions
  • Research or monitoring agents that read arbitrary URLs
  • A reusable pattern for putting a governance proxy in front of any MCP tool

Setup

Connect an LLM credential in the Agent node and keep the policy on browser_agent.

The Agent's MCP connection is preconfigured (uvx arc-gate-mcp wrapping uvx mcp-server-fetch); Heym launches it on run, so there is no manual install step. The PyPI package is at pypi.org/project/arc-gate-mcp.

Notes

Arc Gate MCP is a third-party governance proxy and was released recently, so review its policy and data handling before production use - it proxies all fetched content. To run the Fetch MCP directly without governance, point the Agent's MCP connection at uvx mcp-server-fetch.

How to import this template

  1. 1Click Import → Copy JSON on this page.
  2. 2Open your Heym and navigate to a workflow canvas.
  3. 3Press Cmd+V / Ctrl+V — nodes appear instantly.
  4. 4Add your API keys in the node config panels and click Run.
More workflow templates
View all templates
Heym
incident analysis · production AI
Observed across 100s of AI rollouts

AI workflows don't fail because of prompts.
They fail because of orchestration.

symptom · glue code01
5 tools
Scripts, vector DB, approval bot, tracing, browser runner — none of them talk.
symptom · visibility02
~0%
Observable behavior across the stack. Debugging is guesswork.
with heym · one runtime▲
1 canvas
Agents, RAG, HITL, MCP, traces & evals. Self-hosted. Observable.
AI-Native RuntimeProduction-Grade
github.com/heymrun/heym