root/fusion
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

FN-5833: fix custom Responses API model lookup

Browse Source 
Ensure custom OpenAI Responses providers resolve configured models without requiring registry entries.

- update global settings model resolution to preserve custom provider model configuration
- adjust engine custom-provider and fn-agent creation paths to avoid false 'model not found' failures
- add regression tests for OpenAI Responses custom providers and pi create-fn-agent flows
- document the custom provider model behavior and add a changeset for @runfusion/fusion

Files changed:
 .../FN-5833-custom-provider-responses-fix.md       | 10 ++++
 docs/settings-reference.md                         |  2 +-
 packages/core/src/global-settings.ts               | 56 +++++++++----------
 packages/core/src/index.ts                         |  2 +-
 .../custom-providers-openai-responses.test.ts      | 64 ++++++++++++++++++++++
 .../engine/src/__tests__/custom-providers.test.ts  | 33 ++++++++++-
 .../src/__tests__/pi-create-fn-agent.test.ts       | 11 +++-
 packages/engine/src/custom-providers.ts            |  4 +-
 packages/engine/src/pi.ts                          | 21 +++++--
 9 files changed, 162 insertions(+), 41 deletions(-)

Fusion-Task-Id: FN-5833

Fusion-Task-Lineage: a59481cf-faf5-42b8-97ce-7b3eca27b4fb
This commit is contained in:
gsxdsm
2026-06-01 08:23:02 -07:00
parent b4230c06b1
commit f76716e55b
9 changed files with 167 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
.changeset/FN-5833-custom-provider-responses-fix.md Normal file
View File

@@ -0,0 +1,10 @@
---
"@runfusion/fusion": patch
---
Fix custom-provider model resolution in the bundled engine for OpenAI Responses API providers.
- Align custom-provider reads with global settings directory resolution (including legacy `~/.pi/fusion` and `~/.pi/kb` migration paths), so providers persist across restart and remain visible during agent session creation.
- Ensure custom provider registration diagnostics include enough detail for troubleshooting registration failures.
- Improve configured-model resolution errors to clearly identify the failing `provider/model` selection while retaining the existing `"was not found in the pi model registry"` matcher substring and pointing users to Settings → Custom Providers.
- Add regression tests covering legacy settings-path custom-provider loading and openai-responses provider model resolution.