chore(tests): Mock LLM in tests for PRs

This saves time when testing on CPU which is the only sensible thing
to do on GitHub CI for PRs. For releases or once the commit is merged
we could use an external runner with GPU or just wait.

Signed-off-by: Richard Palethorpe <io@richiejp.com>

core/agent/options.go

@@ -7,6 +7,7 @@ import (
 
 	"github.com/mudler/LocalAGI/core/types"
 	"github.com/sashabaranov/go-openai"
+	"github.com/mudler/LocalAGI/pkg/llm"
 )
 
 type Option func(*options) error
@@ -19,6 +20,7 @@ type llmOptions struct {
 }
 
 type options struct {
+	llmClient llm.LLMClient
 	LLMAPI                                                                                       llmOptions
 	character                                                                                    Character
 	randomIdentityGuidance                                                                       string
@@ -68,6 +70,14 @@ type options struct {
 	lastMessageDuration time.Duration
 }
 
+// WithLLMClient allows injecting a custom LLM client (e.g. for testing)
+func WithLLMClient(client llm.LLMClient) Option {
+	return func(o *options) error {
+		o.llmClient = client
+		return nil
+	}
+}
+
 func (o *options) SeparatedMultimodalModel() bool {
 	return o.LLMAPI.MultimodalModel != "" && o.LLMAPI.Model != o.LLMAPI.MultimodalModel
 }