codex-readiness-integration-test

Automate end-to-end integration testing for agentic loops by executing a multi-stage workflow that validates planning, execution, and scoring. This Skill orchestrates the loop to collect evidence, enforce deterministic checks, and evaluate LLM outcomes for repository-driven workflows.

Codex readiness tests apply deterministic checks for planning-before-code changes, verification-after-code changes, and repo-root path constraints. These sequencing validations ensure the agent follows correct structural phases before applying modifications.

Yes, the integration test workflow operates without external dependencies. It leverages internal scripts and reference components to orchestrate the agentic loop, collect evidence, and run multi-stage scoring natively.

Agentic loop tests fail on repo-root path validation when the agent attempts actions outside the designated repository root directory. The deterministic checks enforce this constraint to ensure safe, scoped execution and provide clear guidance when violations occur.

The best way to validate an agent can propose and execute repository changes safely is running a Codex readiness suite. It orchestrates an agentic loop that enforces planning-before-code and verification-after-code sequencing while collecting evidence for scoring.