DEV Community

#llm

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
Tool Grants Are Firewall Rules: How Excessive Agency Turns Permitted Actions Into Production Incidents

Tool Grants Are Firewall Rules: How Excessive Agency Turns Permitted Actions Into Production Incidents

Comments
4 min read
AI Agents Are eval(userInput) With Extra Steps

AI Agents Are eval(userInput) With Extra Steps

Comments
5 min read
Open-Source LLMs Are Unverified Dependencies: The Model Backdoor Gap npm Already Solved

Open-Source LLMs Are Unverified Dependencies: The Model Backdoor Gap npm Already Solved

Comments
5 min read
RAG Poisoning Is a Supply Chain Attack, Not a Prompt Injection Variant

RAG Poisoning Is a Supply Chain Attack, Not a Prompt Injection Variant

Comments
4 min read
Loops, Cron, and Idempotency: Agents That Operate While You Sleep

Loops, Cron, and Idempotency: Agents That Operate While You Sleep

Comments
4 min read
LLM Covert Channels: How AI Agents Exfiltrate Data in Plain Sight

LLM Covert Channels: How AI Agents Exfiltrate Data in Plain Sight

Comments
5 min read
Prompt Caching Is a Timing Oracle: How the 41-80% Cost Win Becomes Cross-Tenant System Prompt Extraction

Prompt Caching Is a Timing Oracle: How the 41-80% Cost Win Becomes Cross-Tenant System Prompt Extraction

Comments
4 min read
MCP Tool Schema Poisoning: When the Framework Layer Authorizes the Attack

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

Comments
6 min read
OPSEC for Agents: Scope, Caps, Monitoring, and Kill Switches

OPSEC for Agents: Scope, Caps, Monitoring, and Kill Switches

Comments
4 min read
99.7% Rejected in 84ms: Why I Stopped Making the Generator Smarter

99.7% Rejected in 84ms: Why I Stopped Making the Generator Smarter

Comments
7 min read
The Real Cost of LLM APIs in 2026: A Developer's Field Guide

The Real Cost of LLM APIs in 2026: A Developer's Field Guide

Comments 1
3 min read
Regression-testing an agent whose output is never the same twice

Regression-testing an agent whose output is never the same twice

Comments
3 min read
It Fit in Memory and Was Still Unusable — Do the Bandwidth Arithmetic First

It Fit in Memory and Was Still Unusable — Do the Bandwidth Arithmetic First

1
Comments
5 min read
The Week AI Shipped With Locks On: GPT-6 Astra and Claude Fable 5.1

The Week AI Shipped With Locks On: GPT-6 Astra and Claude Fable 5.1

Comments
7 min read
Agent Cost Optimization: Token Accounting, Cache, and Dynamic Dispatching

Agent Cost Optimization: Token Accounting, Cache, and Dynamic Dispatching

Comments
4 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.