AutomationSoftware development
codex-autoresearch
Drive Codex toward measurable code goals while reverting failed experiments automatically.
Drive Codex toward measurable code goals while reverting failed experiments automatically.
Use Codex Autoresearch when a repository task has a numeric outcome, such as fewer test failures, better coverage, lower latency, or a smaller binary. The skill confirms the target, editable scope, metric command, regression guard, and run mode, then lets Codex try one focused change per iteration. Improvements are committed and retained; failed or regressive trials are reverted through Git, with an auditable event history. It can run interactively or in the background, but each experiment requires a clean named Git branch and sufficient access to create and revert commits.
Resource types
Use cases
Platforms
Runtime
Turn an arXiv paper into cited code with explicit uncertainty notes.
Run agent code, builds, hosting, and browser tests inside isolated E2B sandboxes.
Capabilities
Public GitHub facts last synced Jul 14, 2026.
Turn a rough coding idea into a measurable contract for long Codex runs.
Preserve project context, decisions, and next steps across AI coding sessions.