fix(providers): clarify misleading docker image LLM error (#312)

[mrigankad]

Fixes

Fixes #312

Problem

On a fresh setup, creating any flow fails with:

failed to get flow provider: failed to get primary docker image: API returned unexpected status code: 404

Because the message says "primary docker image", users (and several issues — #312, #309, and #203) spend time debugging Docker, image pulls, and the registry. The actual failure is unrelated to Docker.

Root Cause

In NewFlowProvider (backend/pkg/providers/providers.go), the first LLM call during flow creation asks the configured provider to select a primary Docker image:

image, err := prv.Call(ctx, pconfig.OptionsTypeSimple, imageTmpl)
if err != nil {
    return nil, fmt.Errorf("failed to get primary docker image: %w", err)
}

When the configured LLM backend is unsupported or misconfigured, prv.Call returns the provider's HTTP 404. The error is wrapped with wording that points at Docker rather than the LLM provider, so the underlying 404 looks like a Docker registry error. As maintainers have noted on #312/#309, the 404 originates from the LLM backend.

Solution

Reword the wrapped error so the failing component (the LLM call) is identified, while still naming the image-selection step:

failed to select primary docker image via llm call: API returned unexpected status code: 404

One-line message change; no behavioral change.

Testing

cd backend && go build ./pkg/providers/
# -> builds OK

Risk Assessment

Minimal. Only the text of a wrapped error string changes; no control flow, signatures, or behavior are affected. The substring primary docker image is retained, so any log scraping keyed on that phrase keeps matching.

[asdek]

hey @mrigankad

thank you for the PR!