SIGN IN SIGN UP
infiniflow / ragflow UNCLAIMED

RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs

0 0 176 Python

fix: validate custom model inputs (#15200)

### What problem does this PR solve?

Closes #15199.

The add-custom-model endpoint is routed through
`/api/v1/providers/:provider_name/instances/:instance_name/models`, but
the handler previously trusted `provider_name` and `instance_name` from
the JSON body instead of the path target. A request could therefore hit
one provider/instance URL while operating on a different body
provider/instance.

The same handler only rejected `model_types` when the slice was nil. An
empty array passed validation and reached
`ModelProviderService.AddCustomModel`, where `request.ModelTypes[0]`
could panic.

This PR makes the path provider/instance authoritative, rejects
mismatched body values, rejects missing or empty `model_types`, and adds
a service-level guard so direct service callers cannot hit the same
panic path.

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)
B
bitloi committed
ea3a5dba11d015a6cba5312f2cf4bc5ee46f0e9e
Parent: 550bdf2
Committed by GitHub <noreply@github.com> on 5/29/2026, 2:15:01 AM