Codex Begins Encrypting Sub-Agent Instructions
The Decoder · rss · 2026-07-15
The Decoder reports that since early June, OpenAI's Codex has been encrypting instructions sent from the main agent to sub-agents, preventing developers from tracking how internal tasks are divided and delegated.
The article also points out that for the larger GPT-5.6 variants, Sol and Terra, this encryption is mandatory. For developers, this means Codex's internal collaboration process has become invisible. To the outside world, it raises a question: when intermediate instructions between agents are unauditable, debugging, governance, and security investigations become significantly harder.
More from coding & agent
- Soft Clamp cuts tool-call overuse in multi-teacher distillation, from 13.7% to 9.0% — antgroup · 2026-07-21
- Agent harness memory loss and compaction are still a major usability problem — adityaag · 2026-07-21
- SpecJudge runs locally on Ollama to pick the right-sized AI model for your project — jokiruiz · 2026-07-21
- A developer maps out six design rules for CLIs that humans and AI agents can both use — yujiezha · 2026-07-21
- A coding-agent skill that forces ADHD-friendly, answer-first output — ayghri · 2026-07-21
- A set of agent skills for CAD, robotics, and hardware design — earthtojake · 2026-07-21