Simplify LLM protocol (#15366)

In this pull request, we change the zed.dev protocol so that we pass the
raw JSON for the specified provider directly to our server. This avoids
the need to define a protobuf message that's a superset of all these
formats.

@bennetbo: We also changed the settings for available_models under
zed.dev to be a flat format, because the nesting seemed too confusing.
Can you help us upgrade the local provider configuration to be
consistent with this? We do whatever we need to do when parsing the
settings to make this simple for users, even if it's a bit more complex
on our end. We want to use versioning to avoid breaking existing users,
but need to keep making progress.

```json
"zed.dev": {
  "available_models": [
    {
      "provider": "anthropic",
        "name": "some-newly-released-model-we-havent-added",
        "max_tokens": 200000
      }
  ]
}
```

Release Notes:

- N/A

---------

Co-authored-by: Nathan <nathan@zed.dev>

crates/proto/proto/zed.proto

@@ -13,13 +13,6 @@ message Envelope {
     optional uint32 responding_to = 2;
     optional PeerId original_sender_id = 3;
 
-    /*
-        When you are adding a new message type, instead of adding it in semantic order
-        and bumping the message ID's of everything that follows, add it at the end of the
-        file and bump the max number. See this
-        https://github.com/zed-industries/zed/pull/7890#discussion_r1496621823
-
-    */
     oneof payload {
         Hello hello = 4;
         Ack ack = 5;
@@ -201,10 +194,8 @@ message Envelope {
 
         JoinHostedProject join_hosted_project = 164;
 
-        CompleteWithLanguageModel complete_with_language_model = 166;
-        LanguageModelResponse language_model_response = 167;
-        CountTokensWithLanguageModel count_tokens_with_language_model = 168;
-        CountTokensResponse count_tokens_response = 169;
+        QueryLanguageModel query_language_model = 224;
+        QueryLanguageModelResponse query_language_model_response = 225; // current max
         GetCachedEmbeddings get_cached_embeddings = 189;
         GetCachedEmbeddingsResponse get_cached_embeddings_response = 190;
         ComputeEmbeddings compute_embeddings = 191;
@@ -271,10 +262,11 @@ message Envelope {
         UpdateDevServerProject update_dev_server_project = 221;
 
         AddWorktree add_worktree = 222;
-        AddWorktreeResponse add_worktree_response = 223; // current max
+        AddWorktreeResponse add_worktree_response = 223;
     }
 
     reserved 158 to 161;
+    reserved 166 to 169;
 }
 
 // Messages
@@ -2051,94 +2043,32 @@ message SetRoomParticipantRole {
     ChannelRole role = 3;
 }
 
-message CompleteWithLanguageModel {
-    string model = 1;
-    repeated LanguageModelRequestMessage messages = 2;
-    repeated string stop = 3;
-    float temperature = 4;
-    repeated ChatCompletionTool tools = 5;
-    optional string tool_choice = 6;
-}
-
-// A tool presented to the language model for its use
-message ChatCompletionTool {
-    oneof variant {
-        FunctionObject function = 1;
-    }
-
-    message FunctionObject {
-        string name = 1;
-        optional string description = 2;
-        optional string parameters = 3;
-    }
-}
-
-// A message to the language model
-message LanguageModelRequestMessage {
-    LanguageModelRole role = 1;
-    string content = 2;
-    optional string tool_call_id = 3;
-    repeated ToolCall tool_calls = 4;
-}
-
 enum LanguageModelRole {
     LanguageModelUser = 0;
     LanguageModelAssistant = 1;
     LanguageModelSystem = 2;
-    LanguageModelTool = 3;
+    reserved 3;
 }
 
-message LanguageModelResponseMessage {
-    optional LanguageModelRole role = 1;
-    optional string content = 2;
-    repeated ToolCallDelta tool_calls = 3;
+message QueryLanguageModel {
+    LanguageModelProvider provider = 1;
+    LanguageModelRequestKind kind = 2;
+    string request = 3;
 }
 
-// A request to call a tool, by the language model
-message ToolCall {
-    string id = 1;
-
-    oneof variant {
-        FunctionCall function = 2;
-    }
-
-    message FunctionCall {
-        string name = 1;
-        string arguments = 2;
-    }
+enum LanguageModelProvider {
+    Anthropic = 0;
+    OpenAI = 1;
+    Google = 2;
 }
 
-message ToolCallDelta {
-    uint32 index = 1;
-    optional string id = 2;
-
-    oneof variant {
-        FunctionCallDelta function = 3;
-    }
-
-    message FunctionCallDelta {
-        optional string name = 1;
-        optional string arguments = 2;
-    }
+enum LanguageModelRequestKind {
+    Complete = 0;
+    CountTokens = 1;
 }
 
-message LanguageModelResponse {
-    repeated LanguageModelChoiceDelta choices = 1;
-}
-
-message LanguageModelChoiceDelta {
-    uint32 index = 1;
-    LanguageModelResponseMessage delta = 2;
-    optional string finish_reason = 3;
-}
-
-message CountTokensWithLanguageModel {
-    string model = 1;
-    repeated LanguageModelRequestMessage messages = 2;
-}
-
-message CountTokensResponse {
-    uint32 token_count = 1;
+message QueryLanguageModelResponse {
+    string response = 1;
 }
 
 message GetCachedEmbeddings {

crates/proto/src/proto.rs

@@ -203,12 +203,9 @@ messages!(
     (CancelCall, Foreground),
     (ChannelMessageSent, Foreground),
     (ChannelMessageUpdate, Foreground),
-    (CompleteWithLanguageModel, Background),
     (ComputeEmbeddings, Background),
     (ComputeEmbeddingsResponse, Background),
     (CopyProjectEntry, Foreground),
-    (CountTokensWithLanguageModel, Background),
-    (CountTokensResponse, Background),
     (CreateBufferForPeer, Foreground),
     (CreateChannel, Foreground),
     (CreateChannelResponse, Foreground),
@@ -278,7 +275,6 @@ messages!(
     (JoinProjectResponse, Foreground),
     (JoinRoom, Foreground),
     (JoinRoomResponse, Foreground),
-    (LanguageModelResponse, Background),
     (LeaveChannelBuffer, Background),
     (LeaveChannelChat, Foreground),
     (LeaveProject, Foreground),
@@ -298,6 +294,8 @@ messages!(
     (PrepareRename, Background),
     (PrepareRenameResponse, Background),
     (ProjectEntryResponse, Foreground),
+    (QueryLanguageModel, Background),
+    (QueryLanguageModelResponse, Background),
     (RefreshInlayHints, Foreground),
     (RejoinChannelBuffers, Foreground),
     (RejoinChannelBuffersResponse, Foreground),
@@ -412,9 +410,7 @@ request_messages!(
     (Call, Ack),
     (CancelCall, Ack),
     (CopyProjectEntry, ProjectEntryResponse),
-    (CompleteWithLanguageModel, LanguageModelResponse),
     (ComputeEmbeddings, ComputeEmbeddingsResponse),
-    (CountTokensWithLanguageModel, CountTokensResponse),
     (CreateChannel, CreateChannelResponse),
     (CreateProjectEntry, ProjectEntryResponse),
     (CreateRoom, CreateRoomResponse),
@@ -467,6 +463,7 @@ request_messages!(
     (PerformRename, PerformRenameResponse),
     (Ping, Ack),
     (PrepareRename, PrepareRenameResponse),
+    (QueryLanguageModel, QueryLanguageModelResponse),
     (RefreshInlayHints, Ack),
     (RejoinChannelBuffers, RejoinChannelBuffersResponse),
     (RejoinRoom, RejoinRoomResponse),