DEV Community

← All Trends

AI Agent Authorization and Security Flaws

44 posts in this trend in the last 7 days • Active about 5 hours ago
MCP Tool Descriptions Are Instructions, Not Metadata -- and That Is the Exploit
roxdavirox profile
Davi

MCP Tool Descriptions Are Instructions, Not Metadata -- and That Is the Exploit

#ai #llm #mcp #security
1
Comments 1
5 min read
Gating Agent Shell Access: Why Containers Aren't Enough and Approval Loops Break
mech_app_ai profile

Gating Agent Shell Access: Why Containers Aren't Enough and Approval Loops Break

#agents #automation #cybersecurity #security
1
Comments 1
7 min read
Harness Engineering: Designing the Runtime Capability Envelope Around an Agent
shakti_mishra_308e9f36b5d profile

Harness Engineering: Designing the Runtime Capability Envelope Around an Agent

#ai #security #architecture #devops
1
Comments 1
11 min read
How execution boundaries reduce the blast radius of AI agent mistakes
andersonvitaease profile

How execution boundaries reduce the blast radius of AI agent mistakes

#ai #agents #security #mcp
Comments
3 min read
# Agent Firewall v2.5: I Attacked My Own AI Agent Security Boundary
shubhbhangoo profile

# Agent Firewall v2.5: I Attacked My Own AI Agent Security Boundary

#ai #programming #security #agents
Comments
3 min read
AI Agents Are eval(userInput) With Extra Steps
roxdavirox profile
Davi

AI Agents Are eval(userInput) With Extra Steps

#agents #ai #llm #security
Comments
5 min read
OAuth Token Leakage in Agentic AI: When the Context Window Becomes a Credential Store
roxdavirox profile
Davi

OAuth Token Leakage in Agentic AI: When the Context Window Becomes a Credential Store

#ai #cybersecurity #llm #security
Comments
4 min read
Agentic AI Security: Sandboxing LLM Tool Calls in Production
ayinedjimi-consultants profile

Agentic AI Security: Sandboxing LLM Tool Calls in Production

#security #ai #python #devops
Comments
4 min read
AI Agent HTTP Tools Are SSRF by Construction: Network Policy Is the Only Real Control
roxdavirox profile
Davi

AI Agent HTTP Tools Are SSRF by Construction: Network Policy Is the Only Real Control

#agents #ai #cybersecurity #security
Comments 1
5 min read
LLM Agent Sandboxes Have Two Escape Surfaces. Only One Gets CVEs.
roxdavirox profile
Davi

LLM Agent Sandboxes Have Two Escape Surfaces. Only One Gets CVEs.

#agents #llm #mcp #security
Comments 2
5 min read
Shadow Agent Problem
stephen_lincol_dd48ddb8ab profile

Shadow Agent Problem

#agents #ai #cybersecurity #security
1
Comments 4
2 min read
Docker Sandboxes for AI Agents: The Right Way to Isolate Untrusted Code Execution
roxdavirox profile
Davi

Docker Sandboxes for AI Agents: The Right Way to Isolate Untrusted Code Execution

#agents #ai #docker #security
Comments
5 min read
MCP Tool Schema Poisoning: When the Framework Layer Authorizes the Attack
roxdavirox profile
Davi

MCP Tool Schema Poisoning: When the Framework Layer Authorizes the Attack

#cybersecurity #llm #mcp #security
Comments
6 min read
AI Agent Checkpoint Tampering: The State File Is the Attack Surface
roxdavirox profile
Davi

AI Agent Checkpoint Tampering: The State File Is the Attack Surface

#ai #cybersecurity #security #vulnerabilities
Comments
5 min read
JSON, CSV, and YAML Are Not Safe Formats for AI Agents: They Are Attack Vectors
roxdavirox profile
Davi

JSON, CSV, and YAML Are Not Safe Formats for AI Agents: They Are Attack Vectors

#ai #cybersecurity #llm #security
1
Comments 1
5 min read
Specification Gaming Is an Attack Surface, Not an Alignment Footnote
roxdavirox profile
Davi

Specification Gaming Is an Attack Surface, Not an Alignment Footnote

#agents #ai #llm #security
Comments
5 min read
Why command allowlists don't protect your AI coding agent
darkaxiom profile

Why command allowlists don't protect your AI coding agent

#agents #ai #security
Comments
2 min read
The Agent Doesn't Have to Be Malicious for the Action to Be Dangerous
seanchen profile

The Agent Doesn't Have to Be Malicious for the Action to Be Dangerous

#ai #agents #security #opensource
1
Comments
6 min read