Codex CLI
Named by 17 kept items in this workspace.
The items that name it 17
General-purpose agent harness evaluated with four core models.
Coding agent that works on a project: reads and edits code, runs commands, and helps implement or debug a change. The page contrasts it with the OpenAI CLI.
Supported agent integration for Tracekit
OpenAI's command-line coding agent (version 0.157.0 discussed, installed via npm install -g @openai/codex); since 0.157.0 an interactive session auto-starts a persistent background app-server daemon
Agent harness running the Codex agents.
A coding agent CLI that, like Claude Code and Cursor, enforces agent tool-call isolation via OS-level primitives rather than containers.
Harness (version 0.144.6) used by Endor Labs to run identical coding tasks against GPT-6 family models for benchmarking on Azure.
Agent harness used to run the coding tasks with each model; versions 0.144.6 and 0.144.4 were used across the three experiments compared.
Open-source project on GitHub listed among OpenAI's Codex tooling
Conversations from this tool can be imported into mu via mu import.
Coding agent that can use a room's served API
Agent platform where unlazy is installed under ~/.codex/skills/unlazy and invoked via $unlazy.
Official local CLI that Desic Terminal can delegate model requests to instead of requiring a provider API key.
Supported host coding agent; installed via codex plugin marketplace add and plugin add
Agent tool with a documented per-protocol setup guide for routing through Yolorouter.
One of the three CLIs run in parallel to independently red-team a plan.
Tool used to run GPT-5.5 for the control-credibility check comparing untreated Yuj against mini-SWE-agent v2.2.8
In every workspace you can read 37
AI
Security
All technologies in this workspace.