Feature: Add LlamaCpp model discovery API

Add GetAvailableModelsAsync() to LlamaCppVisionClient that:
- Queries llama.cpp /v1/models endpoint for available models
- Filters out multimodal projector models (mmproj-*)
- Tracks loaded/unloaded status for each model
- Returns sorted list with loaded models first

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

MoneyMap/Services/AIVisionClient.cs

@@ -1,5 +1,6 @@
 using System.Text;
 using System.Text.Json;
+using System.Text.Json.Serialization;
 
 namespace MoneyMap.Services
 {
@@ -234,6 +235,44 @@ namespace MoneyMap.Services
             _logger = logger;
         }
 
+        /// <summary>
+        /// Get available models from the llama.cpp server.
+        /// </summary>
+        public async Task<List<LlamaCppModel>> GetAvailableModelsAsync()
+        {
+            var baseUrl = _configuration["LlamaCpp:BaseUrl"] ?? "http://athena.lan:11434";
+
+            try
+            {
+                var response = await _httpClient.GetAsync($"{baseUrl.TrimEnd('/')}/v1/models");
+
+                if (!response.IsSuccessStatusCode)
+                {
+                    _logger.LogWarning("Failed to fetch models: {StatusCode}", response.StatusCode);
+                    return new List<LlamaCppModel>();
+                }
+
+                var json = await response.Content.ReadAsStringAsync();
+                var modelsResponse = JsonSerializer.Deserialize<LlamaCppModelsResponse>(json);
+
+                return modelsResponse?.Data?
+                    .Where(m => !m.Id.StartsWith("mmproj-")) // Filter out multimodal projectors
+                    .Select(m => new LlamaCppModel
+                    {
+                        Id = m.Id,
+                        IsLoaded = m.Status?.Value == "loaded"
+                    })
+                    .OrderByDescending(m => m.IsLoaded)
+                    .ThenBy(m => m.Id)
+                    .ToList() ?? new List<LlamaCppModel>();
+            }
+            catch (Exception ex)
+            {
+                _logger.LogError(ex, "Error fetching models from llama.cpp");
+                return new List<LlamaCppModel>();
+            }
+        }
+
         /// <summary>
         /// Send a text-only prompt to the LLM (no image).
         /// </summary>
@@ -468,4 +507,32 @@ namespace MoneyMap.Services
             return trimmed;
         }
     }
+
+    // Models for llama.cpp /v1/models endpoint
+    public class LlamaCppModel
+    {
+        public string Id { get; set; } = "";
+        public bool IsLoaded { get; set; }
+    }
+
+    public class LlamaCppModelsResponse
+    {
+        [JsonPropertyName("data")]
+        public List<LlamaCppModelData>? Data { get; set; }
+    }
+
+    public class LlamaCppModelData
+    {
+        [JsonPropertyName("id")]
+        public string Id { get; set; } = "";
+
+        [JsonPropertyName("status")]
+        public LlamaCppModelStatus? Status { get; set; }
+    }
+
+    public class LlamaCppModelStatus
+    {
+        [JsonPropertyName("value")]
+        public string? Value { get; set; }
+    }
 }