Skip to content
Navigation menu
Search
Powered by Algolia
Search
Log in
Create account
DEV Community
Close
Agent Safety Engineering Series' Articles
Back to hugginf_expert's Series
When Attackers Bring Their Own Agents: A Defensive Gating Playbook
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 6
When Attackers Bring Their Own Agents: A Defensive Gating Playbook
#
ai
#
agents
#
security
#
llm
Comments
Add Comment
7 min read
Moderation
Indirect Prompt Injection Through Tool Descriptions and Tool Output: How Untrusted Metadata Hijacks Agents
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 6
Indirect Prompt Injection Through Tool Descriptions and Tool Output: How Untrusted Metadata Hijacks Agents
#
ai
#
agents
#
security
#
llm
Comments
Add Comment
7 min read
Moderation
Action Safety for Agents: Deciding What an Agent Is Allowed to Do
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 12
Action Safety for Agents: Deciding What an Agent Is Allowed to Do
#
ai
#
machinelearning
#
llm
#
security
Comments
Add Comment
3 min read
Moderation
Your Agent's Gate Threshold Is a Winner's Curse: Tuning Per-Action Thresholds Without Fooling Yourself
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 19
Your Agent's Gate Threshold Is a Winner's Curse: Tuning Per-Action Thresholds Without Fooling Yourself
#
ai
#
machinelearning
#
security
#
agents
Comments
Add Comment
13 min read
Moderation
A Guardrail That Cuts Attack Success 10x Can Still Be the Wrong Guardrail: Reading AgentDojo Numbers Like an Engineer
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 20
A Guardrail That Cuts Attack Success 10x Can Still Be the Wrong Guardrail: Reading AgentDojo Numbers Like an Engineer
#
ai
#
security
#
agents
#
machinelearning
Comments
Add Comment
11 min read
Moderation
The Model Is Not a Security Boundary: What Two Semantic Kernel CVEs Teach About Gating Tool Arguments
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 21
The Model Is Not a Security Boundary: What Two Semantic Kernel CVEs Teach About Gating Tool Arguments
#
ai
#
python
#
security
#
agents
Comments
2
comments
13 min read
Moderation
Training Where the Data Lives: Why On-Prem Training on Inference NPUs Matters for AI Governance
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 26
Training Where the Data Lives: Why On-Prem Training on Inference NPUs Matters for AI Governance
#
ai
#
machinelearning
#
security
#
privacy
Comments
Add Comment
2 min read
Moderation
When Is Agent Memory Safe Enough to Act On? Scale Verification to the Action, Not the Memory
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Oct 31
When Is Agent Memory Safe Enough to Act On? Scale Verification to the Action, Not the Memory
#
ai
#
python
#
agents
#
security
Comments
Add Comment
13 min read
Moderation
Authenticated Is Not Authorized: What a 9.3 Fail-Open Default in an Agent Mesh Teaches About Tool Gateways
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 2
Authenticated Is Not Authorized: What a 9.3 Fail-Open Default in an Agent Mesh Teaches About Tool Gateways
#
ai
#
programming
#
security
#
mcp
Comments
Add Comment
10 min read
Moderation
A Trailing Dot Beat the Agent's Edit Gate: Check the Path the OS Will Open, Not the String the Model Sent
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 3
A Trailing Dot Beat the Agent's Edit Gate: Check the Path the OS Will Open, Not the String the Model Sent
#
ai
#
programming
#
security
#
mcp
Comments
Add Comment
11 min read
Moderation
The "Read-Only" Fast Path Is Your Weakest Gate: A Newline Let Hidden Commands Through an Agent's Bash Check
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 5
The "Read-Only" Fast Path Is Your Weakest Gate: A Newline Let Hidden Commands Through an Agent's Bash Check
#
ai
#
programming
#
security
#
agents
Comments
Add Comment
12 min read
Moderation
A Valid HTTPS URL Ran Code: What the opencode MCP OAuth Bug Teaches About Validating at the Sink
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 7
A Valid HTTPS URL Ran Code: What the opencode MCP OAuth Bug Teaches About Validating at the Sink
#
ai
#
programming
#
security
#
agents
Comments
Add Comment
12 min read
Moderation
One Check, Four Paths: What the Docker Sandboxes sbx cp Symlink Bug Teaches About Guarding Every Position
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 8
One Check, Four Paths: What the Docker Sandboxes sbx cp Symlink Bug Teaches About Guarding Every Position
#
ai
#
programming
#
security
#
agents
Comments
Add Comment
9 min read
Moderation
The Model Never Ran: Why an Agent's Approval Gate Cannot Trust the Wire Between It and Its LLM
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 9
The Model Never Ran: Why an Agent's Approval Gate Cannot Trust the Wire Between It and Its LLM
#
ai
#
programming
#
security
#
llm
Comments
Add Comment
12 min read
Moderation
The Allowlist Guarded the Destination, Not the Route: A Proxies Parameter Leaked Credentials From an Agent HTTP Tool
hugginf_expert
hugginf_expert
hugginf_expert
Follow
Nov 11
The Allowlist Guarded the Destination, Not the Route: A Proxies Parameter Leaked Credentials From an Agent HTTP Tool
#
ai
#
security
#
python
#
llm
Comments
Add Comment
12 min read
Moderation
We're a place where coders share, stay up-to-date and grow their careers.
Log in
Create account