Codex Web GPT v5.0.5

V5 adds Zero Risk Mode: the full Codex MCP harness still runs through ChatGPT, but the launcher never reads or changes the ChatGPT page and never sends a prompt for you.

V5 highlights

• Zero Risk Mode. The launcher prepares and copies the complete Codex prompt; you paste it into ChatGPT, add any images yourself, select the Codex Zero Risk connector plus the model and effort you want, send it, and confirm Sent. Codex tools, patches, terminal calls, structured output, streaming, and native progress stay available through the same turn-bound harness without browser automation.
• A separate Zero Risk MCP setup. Automatic and Zero Risk modes have independent connector identities, Tunnel IDs, API keys, runtime state, and recovery. Switching modes is transactional and never silently reuses the other mode's private credentials.
• Optional Zero Risk Pro profile. The install menu can add a second ChatGPT Web — Zero Risk Pro row with roughly 250,000 usable tokens before compaction. It is intentionally explicit: select ChatGPT Pro manually for every turn; a Pro account is required, and the larger limit may not work with every effort.
• Manual turns that remain real Codex turns. A neutral request ID binds each ChatGPT response to the exact Codex turn through MCP metadata. Concurrent turns stay isolated; reconnects rejoin the same request; cancellation, completion, retained-chat continuation, and compaction are explicit and fail closed.
• Zero Risk compaction. The active ChatGPT agent produces the checkpoint through the bound MCP control, the old chat is retired, and the compacted Codex continuation starts from a fresh manual chat without exposing private control values in the pasted prompt.
• A clearer launcher flow. First-run onboarding now offers With Automation and Zero Risk, defaults to automation, and keeps the same choice in Settings. The MCP guide includes trimmed, accelerated video walkthroughs with an in-launcher fullscreen viewer.
• Three-language product parity. The new setup, mode selection, manual-turn guidance, diagnostics, and model-profile controls are complete in English, Chinese, and Japanese.

v5.0.5 hotfixes

• Subagents keep their working environment. Fixed V1 children without an assigned agent path, inherited visualization folders, cold fork history, and V2 parent instructions that could fail with missing cwd. Addresses #314.
• Windows resume accepts canonical path variants. Native rollout paths with a Windows namespace prefix now match the same sessions directory without dropping ownership checks. Fixes #376.
• Zero Risk drops stale chat bindings after navigation. A changed page receives full context on the next turn; an active incremental continuation stops explicitly if its conversation is lost. Same-page history updates remain valid. Fixes #377.
• Post-compaction tasks recognize the current environment. Codex can send plugins, instructions, and environment as separate parts of one message. Continuation now reads the environment part correctly while retaining the native thread and permission checks.
• Bigger Context fits the actual messages. Two- or three-part planning now accounts for both tokens and composer characters, including images and instructions carried by the final part. Complete records stay in order; successful multipart compaction no longer fails during usage calculation. Addresses local defects investigated in #346 and #353.
• Retained responses survive ChatGPT history re-rendering. Response tracking uses stable turn IDs and accounts for virtualized history, preventing display-index changes from breaking the binding or making old messages look new. Addresses one confirmed failure path in #351.
• Think uses /think. Think is selected after the connector and its state is verified on fresh and retained chats. An already enabled mode stays enabled; switching back to Luna clears it. Go/Windows Full-mode retesting remains tracked in #357.
• MCP readiness is verified before monitoring starts. New and already running tunnels must expose healthy MCP diagnostics. Missing diagnostics can no longer be replaced by an inventory ready flag. Fixes the launcher monitoring gap tracked in #372.
• Codex config edits and symlinks survive setup. A relocated managed-hook comment no longer invalidates unchanged hook definitions. Setup, update, removal, and rollback preserve a symlinked config.toml and its target. Fixes #348 and #352.
• Native Image Gen reaches the backend. Added forwarding for Codex's image generation and editing routes, fixing the local 404 before the generator was reached. Native account authorization and image limits still apply.
• Cleaner streaming and failure diagnostics. Chart loading controls no longer contaminate the text consistency check. Accepted compaction is recorded as completed; lost connector input is identified separately from personalization, native compact failures include route evidence, and the stalled-response diagnostic checkpoint moves to 60 seconds. These changes do not claim to resolve every remaining #321, #323, or #351 report.

Thanks to @b-y-f (#340) and @Evanlau1798 (#347) for the subagent fixes, and to @MaxKerkula (#362) and @DJNgoma (#354) for the continuation and monitoring investigations.

Please update through the launcher, restart Codex, and retry the affected workflow. If the same failure remains, reply in its issue with your versions and a fresh Activity safe-log export from that attempt.

Install

macOS or Linux:

curl -fsSL https://github.com/miuuyy/codex-chatgpt-web/releases/latest/download/install-launcher.sh | sh

Windows PowerShell:

irm https://github.com/miuuyy/codex-chatgpt-web/releases/latest/download/install-launcher.ps1 | iex