What is OpenAI's Codex Skills feature and what are Automations?

Two higher-level Codex product features extend beyond direct coding assistance:

Skills are reusable, project-specific capabilities that Codex learns and applies consistently. They go beyond writing code to encompass code understanding, prototyping, and documentation - aligned with your team's specific standards and patterns. Skills allow Codex to understand your team's conventions, architectural patterns, and tooling preferences, producing outputs that fit directly into your existing workflows.

Automations allow Codex to work unprompted on routine but important tasks. Instead of waiting for a developer to ask, Codex proactively picks up work like:

• Issue triage - categorising and labelling new GitHub issues
• Alert monitoring - responding to CI failures or monitoring alerts
• CI/CD pipeline tasks - running checks, updating dependencies
• Scheduled code quality tasks - running linters, generating reports

# Automations can be configured to:
# - Monitor GitHub issue queues and triage new issues
# - Watch CI/CD pipelines and fix recurring failures
# - Respond to monitoring alerts and attempt automated remediation
# - Run scheduled code quality reviews

# Skills examples:
# - "Our team uses pytest with the Arrange-Act-Assert pattern"
# - "All new API endpoints need OpenAPI docstrings following our schema"
# - "Refactoring should preserve backward compatibility per our versioning policy"

# These are configured in the Codex App UI, not via the API directly

Together, Skills and Automations represent the shift from Codex as a passive responder to an active software engineering team member that proactively contributes to quality and productivity without requiring a developer to initiate every interaction.