deep-agents-core

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFE
Full Analysis
  • Filesystem Operations: The skill describes tools for managing files, such as read_file, write_file, and edit_file. While these are central to the framework's context management, they represent a powerful capability that allows agents to interact with the local environment.
  • Task Delegation (SubAgents): The framework supports spawning specialized subagents to handle subtasks. This capability is designed for modular task management and allows for scaling complex operations by delegating work to autonomous components.
  • Dynamic Instruction Loading: The skill outlines a mechanism for on-demand loading of specialized capabilities from a directory. This architecture enables agents to adapt their instructions and tools based on the specific requirements of a task.
  • Data Ingestion Surface: As a framework that processes external files and user-provided messages, it creates a surface where instructions embedded in data could influence agent behavior. The framework includes middleware options like human-in-the-loop approvals to help manage these interactions safely.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 01:44 PM