TL;DR: "Claude can't draw" is a budget problem, not a capability problem. codex-bridge (MIT) is a Claude Code plugin that routes image generation (gpt-image-2) and five GPT-5 subagents through the Codex CLI login you already have — billing against your ChatGPT plan instead of a second API key. I have not tested or run it; this is a source-reading review.
How it works. Two bash wrappers shell out to codex exec
, Codex CLI's non-interactive mode, with your existing codex login
(both scripts check codex login status
first). Text tasks use a read-only
sandbox and capture only the final answer via -o <tmpfile>
(`bin/codex-run:92-96`
). Image tasks use a `workspace-write`
sandbox scoped to the output directory (bin/codex-imagegen:110
), then verify the file landed. Exit codes: 0 ok, 1 precondition, 2 failure, 124 timeout.
Why the split matters. Codex runs as a separate process, so its intermediate output never enters your Claude context — you pay Claude tokens only for the brief and the review. The trade-off: Codex is blind to your conversation history, so every brief must stand alone, which is why every subagent is instructed to verify before reporting.
The delegation rubric (skills/codex-delegate/SKILL.md
) is the most reusable part: delegate high-volume, low-judgment work (mechanical multi-file edits, bulk scaffolding, exhaustive sweeps, all image work, long self-contained subtasks); keep low-volume, high-judgment work (architecture calls, ambiguous requirements, anything needing conversation history, edits under ~3 files). Round-tripping a two-line fix costs more, not less.
Trust boundary. No API key: credentials live in ~/.codex
from codex login
, and every call forwards your task text to OpenAI's servers. If your team's data policy cannot accept a second vendor seeing prompts, that is a blocker, not a footnote.
Caveats from README and source. 1–4 minutes per image; image turns burn ChatGPT quota ~3–5× faster than text; the default path cannot emit native alpha (chroma-key strip, or CLI fallback plus API key); macOS and Linux only; custom sizes must satisfy longest edge ≤3840px, both edges multiples of 16, ratio ≤3:1, and 655,360–8,294,400 total pixels.
Who should try it. You already pay for a ChatGPT plan, run Claude Code daily, and regularly produce images or mechanical bulk edits. Skip it if you rarely need images, your work is small and interactive, or your data rules forbid OpenAI seeing prompts.
Project: https://github.com/Sateezg/codex-bridge · README: https://github.com/Sateezg/codex-bridge/blob/main/README.md · codex-run: https://github.com/Sateezg/codex-bridge/blob/main/bin/codex-run · codex-imagegen: https://github.com/Sateezg/codex-bridge/blob/main/bin/codex-imagegen · delegate rubric: https://github.com/Sateezg/codex-bridge/blob/main/skills/codex-delegate/SKILL.md