Urgent.News

What's breaking now, across thousands of outlets.

AI

Does a LangGraph subagent inherit its parent's tool permissions? I tested it.

I gave a supervisor agent in Deep Agents 0.7.6 exactly one tool, write_brief . It spawned a writer subagent. The writer ran a web search for site:exfil.example internal customer list , because a note in its input told it to. Nothing checked that call against what the supervisor held. This isn't a bug. LangChain's Deep Agents permissions page says a subagent spec "replaces the parent's rules…

A LangGraph subagent does not inherit its parent's tool permissions by default. In the testing conducted, a supervisor agent with write_brief permission spawned a writer subagent that ran a web search. The subagent executed the web search despite lacking write_brief permission and there was no mechanism to restrict it. Deep Agents documentation states that subagents replace the parent's rules entirely with their own tool sets.

To bound a subagent, a middleware named BoundedByParent can be added to its spec which checks that the subagent can only use tools held by the parent. Attenu-guard library extends this functionality with scopes, ceilings and lifetimes for finer control of permissions.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

I tried to build an AI anime on a 16GB MacBook. Here is exactly where it broke

I wanted to find out whether a laptop can make an animated short with AI. Not "can a model generate a picture" — the whole thing: music, voices, character art that stays the same person across shots…

  • 16GB MacBook struggled to generate video content
  • Music synthesis succeeded with custom solution
  • Voice synthesis required careful data quality

Aegisora 2.0: A Runtime Security Layer for Autonomous AI Agents

AI agents are moving from generating text to taking actions — calling APIs, using tools, accessing data, and executing workflows. That changes the security problem.

  • Aegisora 2.0 secures autonomous AI agents' actions, not just model output
  • System intercepts, analyzes, enforces policies before external system access
  • Open-source tool for developers managing agents with real tool/API access

More from Wednesday 2 September →