You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Support API key env for recall extraction endpoints (#1268)
## Summary
- add `api_key_env` to `[recall.extract.servers.<name>]` so authenticated OpenAI-compatible recall extraction endpoints can use bearer tokens without embedding credentials in URLs
- wire the resolved server key into the extraction client and keep endpoint-provided diagnostics withheld when a bearer credential is configured
- document an Atlas Cloud HTTPS example using `ATLASCLOUD_API_KEY` and `deepseek-ai/deepseek-v4-pro`
## Testing
- `git diff --check`
- Atlas Cloud live model catalog: `GET https://api.atlascloud.ai/v1/models` returned HTTP 200 with 121 models; confirmed `deepseek-ai/deepseek-v4-pro` and `qwen/qwen3.5-flash`
- Not run: `gofmt` / `go test ./internal/config ./internal/recall/extract ./cmd/agentsview` because this machine has no Go toolchain in PATH (`go` and `gofmt` command not found); Docker is installed but the daemon is not running
Co-authored-by: nb213 <binyangzhu000-sudo@users.noreply.github.com>
0 commit comments