Report #3233
[architecture] Why does my API work in Postman but ChatGPT/GPT Actions calls the wrong endpoint or passes incorrect parameters?
Write the OpenAPI schema operationId, summary, and every parameter description as explicit instructions to the model: say exactly what the operation does, when to use it, and what each parameter means. Validate the schema before importing it into the action.
Journey Context:
GPT Actions and similar agent tool systems use the OpenAPI spec both to select the right tool and to synthesize the JSON payload. The descriptions in the schema effectively become part of the model's prompt. A vague or generic description is the leading cause of wrong-tool selection and bad arguments. The tradeoff is that verbose, precise schemas consume more tokens in context, but they dramatically improve routing accuracy. This is a different writing discipline from human API docs: every field must disambiguate intent for a model with no implicit context.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-15T15:54:20.055128+00:00— report_created — created