Open Codex repeats identical tool calls instead of bounding the loop

Summary

Severity / performance impact: High. A repeated tool-use response can cause repeated provider turns and duplicated tool-result submission until timeout.

Affected target

• Target: Open Codex
• Version/build: open-codex 0.1.31
• Environment: Linux x86_64 Docker, public package build
• Provider/model: OpenAI-compatible local mock provider, gpt-4

Steps to reproduce

Prerequisites: Docker, Python 3, and the GitHub CLI (gh) for the clone command below. The linked reproducer is self-contained and uses only Python standard-library modules plus Docker. It builds the affected CLI version from the public package registry, starts a local mock provider, and runs the CLI in an isolated workspace. The Docker run is limited to 2 CPUs and 4 GiB RAM by default.

Complete self-contained reproducer: https://gist.github.com/N0zoM1z0/1c46e55fa8d28f83ed82db06f3cbd101

The Gist contains open-codex-repeated-identical-tool-loop.reproduce.py. Download and run:

gh gist clone 1c46e55fa8d28f83ed82db06f3cbd101 open-codex-repeated-identical-tool-loop-reproducer
cd open-codex-repeated-identical-tool-loop-reproducer
python3 open-codex-repeated-identical-tool-loop.reproduce.py

To reuse an already-built local image:

python3 open-codex-repeated-identical-tool-loop.reproduce.py --skip-build

Expected successful reproduction output includes:

REPRODUCED

Actual behavior

The local provider repeats a shell tool call after tool results arrive. The script checks for repeated tool-result submission, provider requests, and timeout/signal termination.

Expected behavior

Open Codex should detect repeated identical tool-use state and stop with a bounded error rather than resubmitting the same tool result repeatedly.

Public-upstream status

This reproduces against stock public open-codex@0.1.31 built from npm.