Learn about this first
OpenRouter has published a tutorial explaining how to build an AI code-review script with its Responses API: a 233-line review.py file using only the standard library uploads changed files to a hosted Linux sandbox, allows the model to actually run checks to confirm bugs, and returns fixes in V4A patch format.
The tutorial says it requires Python 3.11 or later, Git, and an OpenRouter API key. The Files API, Shell tool, and Apply Patch tool used are available only at the global endpoint (https://openrouter.ai/api/v1); regional endpoints return 403. All three are in beta, so their behavior may change. The Shell tool accepts at most 20 file IDs per request (19 changed and context files plus the diff).
AI-generated from reports · updated 16 days agoLatest developments
OpenRouter tutorial: Build an AI code review script that verifies bugs in a Linux sandbox using the Responses API
Event progress1
OpenRouter Publishes Tutorial on Building an AI Code Reviewer
OpenRouter: Announcements: OpenRouter Tutorial: Build an AI Code-Review Script That Can Verify Bugs in a Linux Sandbox with the Responses API