</> HitReader
Blog Explore About

Tag: #observability

When Sub-Agent Prompts Go Dark: Encrypted MultiAgentV2 in Codex
agentic ai & tooling Jul 14, 2026 6 min read

When Sub-Agent Prompts Go Dark: Encrypted MultiAgentV2 in Codex

Codex’s MultiAgentV2 started encrypting sub-agent `message` payloads, storing only ciphertext (`encrypted_content`) and leaving plaintext empty (`content`). That improves privacy, but can erase the readable delegation details in rollout/history, making audits and debugging harder.

by ahsan
#agentic ai security #codex #encryption #multi-agent #observability
Why Some Coding Agents Burn 33k Tokens Before You Type Anything
llm engineering Jul 12, 2026 6 min read

Why Some Coding Agents Burn 33k Tokens Before You Type Anything

Some coding agents arrive at the model already “loaded” with tens of thousands of tokens of system prompt and tool scaffolding. A side-by-side comparison found ~33k-token overhead for Claude Code versus ~7k for OpenCode, with prompt caching benefits depending on prefix stability.

by ahsan
#engineering #llm agents #observability #prompt caching #token cost

Categories

  • machine learning
  • software engineering
  • cybersecurity
  • artificial intelligence
  • embedded systems
  • systems programming
  • web development
  • llm engineering
Explore all →

Tags

#llm #privacy #open-source #ai #rust #cybersecurity #linux #security #ai agents #machine learning #web development #javascript
Explore all →

© 2026 HitReader.

About Explore Terms Privacy Facebook