OpenAI Codex Team Fixes OpenClaw Authentication Bug, Significantly Improves Agent Behavior

Gate News message, April 23 — The OpenAI Codex team is focused on optimizing the OpenAI model experience in OpenClaw, with Codex engineering lead Tibo Sottiaux collaborating with OpenClaw creator Peter Steinberger. Codex product lead Nik Pash discovered a critical authentication flaw: when OpenClaw was configured to use Codex harness with OpenAI models, the authentication process failed and the system silently fell back to Pi harness, causing users to believe Codex harness was functioning normally when it was not.

Pash submitted two pull requests to address the issue: one to fix the authentication bridge and another to prevent silent fallback. The improvements stem entirely from switching the underlying runtime adapter (harness) that governs how OpenClaw communicates with the model API, while the agent’s prompt and higher-level workflow logic remained unchanged.

Agent behavior showed marked differences before and after the fix. With Pi harness, the agent performed shallow polling on each heartbeat: reading heartbeat files, checking Discord, returning HEARTBEAT_OK, and ignoring other instructions. It sometimes inferred operations to execute but failed to issue tool calls. After switching to Codex harness, the agent entered a full work loop: reading workspace context, parsing task lists, checking repositories, executing edits, and attempting verification. Subsequent heartbeats could resume progress rather than repeat work.

Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to Disclaimer.

Related Articles

AWS, Coinbase, Stripe Launch USDC Payments for AI Agents

Amazon Web Services announced Thursday the launch of "Amazon Bedrock AgentCore Payments," a new set of features enabling AI agents to execute transactions using stablecoins in partnership with Coinbase and Stripe. The platform allows AI agents to instantly access and pay for services including web c

CryptoFrontier2h ago

CZ Says YZi Labs Allocates 70% to Blockchain, 20% to AI at Consensus Miami 2026

According to ChainCatcher, at Consensus Miami 2026, Zhao Changpeng (CZ) stated that YZi Labs allocates 70% of funding to blockchain, 20% to AI, and 10% to biotechnology. CZ added that BNB should be positioned as a native currency for AI agents, and all blockchains need to be "AI ready" to support

GateNews4h ago

Zypher Network Integrates AIDEN to Improve Blockchain Knowledge Accessibility

According to IQ AI's official announcement on May 6, Zypher Network integrated AIDEN, an artificial intelligence agent developed by IQ AI, to improve blockchain knowledge accessibility and strengthen auditability within AI-powered ecosystems. The collaboration introduces a dedicated wiki page

GateNews4h ago

Virtuals Protocol Launches OpenGradient Titan Airdrop, Distributes 500K OPG Today

According to Virtuals Protocol official announcement, the OpenGradient Titan Launch airdrop is now live as of May 7. Eligible users can claim OPG tokens directly from their Virtuals accounts. A total of 500,000 OPG tokens are being distributed today to reward contributors to the Virtuals

GateNews7h ago

NVIDIA releases Nemotron 3 Nano Omni open-source multimodal model

NVIDIA releases Nemotron 3 Nano Omni, an open-source multimodal model that integrates vision, speech, and language into a single model as the perception layer for AI agents. The core is a 30B-A3B MoE with a 256K context and 9x throughput. It supports inputs including text, images, audio, and video, and outputs text. In conjunction with Nemotron 3 Super/Ultra, it is suitable for scenarios such as computer operations, document intelligence, and audio-visual understanding, and it publishes weights and training data for deployment spanning local environments, NIM, and third-party platforms.

ChainNewsAbmedia8h ago

NeoSoul and AllScale Announce Strategic Partnership on Agent Credit and Stablecoin Settlement Today

According to ChainCatcher, AI Agent Economy project NeoSoul announced a strategic partnership with AllScale today (May 7) to explore credit formation and stablecoin settlement mechanisms for autonomous Agent collaboration. The partnership will focus on automating payment flows between Agents—such

GateNews9h ago
Comment
0/400
No comments