Category field guide
Repository-aware coding assistants and agents used to investigate, plan, edit, test, and return software changes for human review.
My position
GitHub Copilot, OpenAI Codex, and Jules are not interchangeable model names. Copilot is a GitHub-centered workflow that can expose multiple model providers; Codex and Jules can also operate as direct delegated agents with their own environments, permissions, and review paths.
The primary control boundary is the repository and execution context: organization, repo, branch, working directory, credentials, network access, commands, changed files, tests, and the human approval required before merge or deployment.
The working set
A field-note label is not a downgrade or a placeholder. It is the honest boundary between useful experience and a completed repeatable review.
The primary GitHub-centered coding profile, spanning IDE assistance, repository-aware agent work, pull requests, and selectable model providers.
Use Copilot as the workflow surface; evaluate Claude, Codex, and Gemini as model coverage inside that governed repository workflow.Read the profile →OpenAI's direct software-engineering agent for reading repositories, editing files, running commands and tests, and returning reviewable work across app, CLI, IDE, and cloud surfaces.
Use Codex directly when the task needs an accountable engineering agent outside Copilot; never let it guess the repository or deployment target.Read the profile →Google's GitHub-connected coding agent for planned repository work in a remote virtual machine, including bugs, features, tests, and documentation.
Evaluate Jules as a delegated GitHub task agent, with plan approval and branch review treated as mandatory controls.Read the profile →Selection rules
A reviewable software change with a known scope, reproducible validation, and no ambiguity about where the agent operated.
