The Pulse
Qodo Brings Company Context to OpenAI Codex Reviews
Qodo’s Codex plugin brings company rules and codebase context into coding sessions, then returns independent review findings before a developer opens a pull request. OpenAI’s profile also describes a possible use of its Decisions API to sel
AI.info Team · Written with AI · fact-checked ·
Codex can finish a local coding task efficiently, but a change that looks correct in one repository may break a service elsewhere. Qodo’s new Codex plugin is built around that tension: it brings company rules and codebase context into the coding session, then sends the resulting changes to a separate review agent. The aim, as described in an OpenAI profile published November 2, is to catch problems that a local view can miss without removing developers from the decision-making process.
Qodo builds AI tools for code review and software quality, giving coding agents organizational knowledge to understand the code they change. Its Codex plugin brings that context and review into developers’ coding sessions.
“We love OpenAI models. We want the frontier, the best reasoning models, but also those that are really fast.”
Itamar Friedman, co-founder and CEO, Qodo
When a change crosses repository boundaries
Friedman describes customer environments with 10,000 or more repositories. In that setting, a code change can appear sound in the repository being edited and still cause trouble for a downstream service. Dependencies may also cross different Git providers or connect through services such as Kafka, making it harder for a coding agent working on one task to see the wider system.
Qodo’s plugin, available in the Codex marketplace, gives Codex access to codebase knowledge, organizational rules and Qodo’s Agentic Toolbox for independent review. The OpenAI profile presents that cross-repository view as a way to help Codex identify serious issues, including what it calls P0 bugs. It does not provide a measured detection rate or a comparison against Codex without the plugin.
Rules first, review before the pull request
The workflow can begin before Codex writes code. Qodo’s plugin can retrieve relevant codebase rules and pull-request history and patterns, giving the coding agent company-specific guidance for its task. After Codex makes a change, Qodo reviews it and returns findings to Codex in the same session, before the developer opens a pull request.
That arrangement separates code generation from review rather than asking one agent to do both jobs. Qodo’s AI Developer Relations Lead, Nnenna Ndukwe, says developers are using the workflow to build features and address technical debt with GPT-6 models in Codex. The OpenAI profile describes these as observed workflows, not a published evaluation of review quality.
Qodo wants developers to judge the disagreement
Friedman calls coding and review agents “yin and yang”: one produces code while the other challenges it. His stated direction is for agents to discuss findings and surface the issues they cannot resolve, leaving developers to examine the remaining disagreements alongside the relevant context and earlier decisions.
That model keeps human judgment in the loop, but changes where it is applied. Rather than reviewing every step of code production, developers would focus on unresolved questions raised by the agents. The profile describes that as a vision for the workflow, not a feature that has already been demonstrated as fully automated.
The Decisions API could narrow the rules
Friedman also points to OpenAI’s Decisions API as a possible way to choose which rules and agent skills apply at different stages of development. He gives the example of selecting from 200 rules before asking a language model to reason through the relevant ones, which he says could make selection faster and cheaper while improving review quality.
Qodo’s plugin today is described as supplying company context and independent review within Codex. The Decisions API example is a proposed use, not a reported capability of the plugin. OpenAI’s account offers no cost figures, measured review gains or independent test results; its concrete description is a sequence of steps that returns Qodo’s findings to Codex before a developer opens a pull request.