I cant able to use requesty key with the docker command

[manku211]

docker run --rm -it
-e OPENAI__KEY=sk-XXXXXXX
-e OPENAI__API_BASE=https://router.requesty.ai/v1
-e CONFIG__MODEL=gpt-4.1
-e CONFIG__FALLBACK_MODELS='["gpt-4.1"]'
-e GITHUB__USER_TOKEN=ghpXXXXXXXXX
codiumai/pr-agent:latest
--pr_url https://github.com/XXXXXXXXX

this is my command , do I need to use locally with python command instead to change the model

[IsmaelMartinez]

Two things, and neither needs a local Python install. First, the command is missing the tool: --pr_url on its own fails argument parsing, so append review (or describe, improve) as the last argument, as the local installation examples show. Second, this fork publishes pragent/pr-agent:latest, so use that rather than the old codiumai image. The environment variables themselves are close to right, since OPENAI__KEY and OPENAI__API_BASE are forwarded on every call; if the router expects a namespaced model id, keep the openai/ prefix on it and add config.custom_model_max_tokens, which is the pattern documented for the other OpenAI-compatible gateways in the changing-a-model guide. Closing as answered given the age; please open a fresh discussion if it still fails after those corrections.