Frequent 429 Rate Limit errors when using Codex with Spec-kit for automated development.
Resolved 💬 3 comments Opened May 1, 2026 by aaronchen2k Closed May 1, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
@dosu
What issue are you seeing?
I am currently using Codex combined with Spec-kit for automated project development. However, I frequently encounter HTTP 429 (Too Many Requests) errors during the workflow. This interrupts the automation process and prevents Spec-kit from completing the task chain (e.g., during /speckit.specify or /speckit.tasks execution).
What steps can reproduce the bug?
Initialize a project using Spec-kit with Codex as the AI agent (specify init my-project --ai codex).
Run a Spec-kit command, such as /speckit.specify or /speckit.plan, to generate specifications.
Wait for Codex to process the request.
See error: The process fails with a 429 Rate Limit error (often related to RPM or TPM limits).
What is the expected behavior?
_No response_
Additional information
_No response_
3 Comments
Are you using an OpenAI endpoint or Azure?
@etraut-openai use Azure
Please report the issue to the Azure team. This is probably because you're running short on quota, and you'll need to adjust your TPM limits.