Releases·Agent Release
OpenAI gives Codex reusable cloud environments that work across devices
At Dev Day, OpenAI announced reusable cloud development environments for Codex, accessible from any device, plus a refreshed Codex CLI with voice control, a new code review experience, Codex Security Cloud, and API updates. The changes make Codex cloud work persistent and configurable rather than isolated per-task sandboxes.
4121 engagement2 sourcesTue, Sep 29, 2026, 05:15 PM
AI Overview
- Codex gains persistent, configurable cloud environments accessible from any device, announced at OpenAI's Dev Day.
- Updated Codex CLI adds voice task direction, a /agents view for delegating multiple tasks, and a cleaner terminal UI.
- New code review in the ChatGPT desktop app lets Codex review GitHub PRs and GitLab MRs, including automatic reviews.
- Codex Security Cloud scans GitHub repositories on demand or schedule and prepares fixes using models from OpenAI's Daybreak Blue initiative.
- API updates include a Decisions API powered by Luna and an Agents API supporting computer use and Amazon Bedrock Managed Agents on AWS.
Related
Releases·Tool Release
OpenAI ships Codex 0.151.0 with MCP tool grace period and extension hooks
OpenAI released Codex 0.151.0 on GitHub on 2026-08-29, following alpha builds 0.151.0-alpha.6 and alpha.7.2. The stable release adds a configurable grace period for discovering tools from optional MCP servers, lets extensions inspect or replace MCP tool results before they reach the model, and fixes sandbox and permission bugs. It matters to developers running Codex CLI with MCP servers and plugin catalogs.
OpenAI released Codex 0.151.0 on GitHub on 2026-08-29, following alpha builds 0.151.0-alpha.6 and alpha.7.2. The stable release adds a configurable grace period for discovering tools from optional MCP servers, lets extensions inspect or replace MCP tool results before they reach the model, and fixes sandbox and permission bugs. It matters to developers running Codex CLI with MCP servers and plugin catalogs.
Updated Sun, Aug 30, 2026, 01:56 PM — OpenAI released Codex 0.152.0-alpha.4 on 2026-08-30, a day after the stable 0.151.0 release.
04022 engagement4 sourcesgithubSat, Aug 29, 2026, 12:00 AM
Social·Community
12 hours, 4,518 lines of tooling it built for itself, and the app still loads in 35 seconds (GPT-6.1 Sol case report)
A solo developer documented a failed 12-hour Codex CLI run using gpt-6.1-sol (reasoning effort 'high') meant to speed up a 35-second-loading internal reporting app. The agent never measured the actual load phases, instead writing 4,518 lines of analysis tooling plus an 8,183-line test file, interrupting its own subagent, and integrating nothing. The developer cancelled their $200/month subscription and filed an issue at openai/codex#502.
A solo developer documented a failed 12-hour Codex CLI run using gpt-6.1-sol (reasoning effort 'high') meant to speed up a 35-second-loading internal reporting app. The agent never measured the actual load phases, instead writing 4,518 lines of analysis tooling plus an 8,183-line test file, interrupting its own subagent, and integrating nothing. The developer cancelled their $200/month subscription and filed an issue at openai/codex#502.
14 engagement1 sourceredditFri, Oct 2, 2026, 11:47 AM
Releases·Tool Release
OpenAI Codex CLI v0.144.0 adds usage-limit credit details, writes approval mode, and interactive MCP auth
OpenAI released Codex CLI v0.144.0 on July 9, 2026, introducing usage-limit reset credits with type/expiration display and selection, a new `writes` app-approval mode for granular read/write permissions, and interactive authentication for MCP tools without experimental opt-in. These features improve developer control over API usage and security.
OpenAI released Codex CLI v0.144.0 on July 9, 2026, introducing usage-limit reset credits with type/expiration display and selection, a new `writes` app-approval mode for granular read/write permissions, and interactive authentication for MCP tools without experimental opt-in. These features improve developer control over API usage and security.
14 engagement1 sourcegithubThu, Jul 9, 2026, 04:47 PM
Releases·Tool Release
ai-memory-skillpack: fail-closed project-memory skill pack for Codex CLI ships as MIT-licensed v1.25.1 on GitHub
The ai-memory-skillpack project shipped a bounded, fail-closed project-memory system positioned as 'the memory plane for Codex', targeting OpenAI's Codex CLI with optional support for Claude Code from the same skill source. It gives practitioners a way to add persistent project memory to coding agents without unbounded memory growth.
The ai-memory-skillpack project shipped a bounded, fail-closed project-memory system positioned as 'the memory plane for Codex', targeting OpenAI's Codex CLI with optional support for Claude Code from the same skill source. It gives practitioners a way to add persistent project memory to coding agents without unbounded memory growth.
24 engagement1 sourcegithubTue, Sep 8, 2026, 05:07 AM
Releases·Tool Release
Developer builds Baton CLI to transfer sessions between 9 AI coding agents
A developer created Baton, an open-source Rust CLI that converts session files between nine coding agents (Codex, opencode, Gemini CLI, Cursor, Zed, Aider, Cline, Continue, Claude Code), allowing users to resume conversations after hitting usage limits without re-explaining context. The tool was itself mostly written by Claude Code, including the parser for its own session format.
A developer created Baton, an open-source Rust CLI that converts session files between nine coding agents (Codex, opencode, Gemini CLI, Cursor, Zed, Aider, Cline, Continue, Claude Code), allowing users to resume conversations after hitting usage limits without re-explaining context. The tool was itself mostly written by Claude Code, including the parser for its own session format.
5 engagement2 sourcesredditSun, Jul 12, 2026, 08:26 PM