model: gemini integration (#99)

* Integrate gemini

* Fix gemini cannot read current dir content

* Update June devlog

* Prepare for release 0.1.3

* Add dev log for july

Author: honganh1206

.zed/tasks.json

@@ -0,0 +1,47 @@
+[
+  {
+    "label": "Run server",
+    "command": "make serve",
+    //"args": [],
+    // Env overrides for the command, will be appended to the terminal's environment from the settings.
+    "env": { "foo": "bar" },
+    // Current working directory to spawn the command into, defaults to current project root.
+    //"cwd": "/path/to/working/directory",
+    // Whether to use a new terminal tab or reuse the existing one to spawn the process, defaults to `false`.
+    "use_new_terminal": false,
+    // Whether to allow multiple instances of the same task to be run, or rather wait for the existing ones to finish, defaults to `false`.
+    "allow_concurrent_runs": false,
+    // What to do with the terminal pane and tab, after the command was started:
+    // * `always` — always show the task's pane, and focus the corresponding tab in it (default)
+    // * `no_focus` — always show the task's pane, add the task's tab in it, but don't focus it
+    // * `never` — do not alter focus, but still add/reuse the task's tab in its pane
+    "reveal": "always",
+    // Where to place the task's terminal item after starting the task:
+    // * `dock` — in the terminal dock, "regular" terminal items' place (default)
+    // * `center` — in the central pane group, "main" editor area
+    "reveal_target": "dock",
+    // What to do with the terminal pane and tab, after the command had finished:
+    // * `never` — Do nothing when the command finishes (default)
+    // * `always` — always hide the terminal tab, hide the pane also if it was the last tab in it
+    // * `on_success` — hide the terminal tab on task success only, otherwise behaves similar to `always`
+    "hide": "never",
+    // Which shell to use when running a task inside the terminal.
+    // May take 3 values:
+    // 1. (default) Use the system's default terminal configuration in /etc/passwd
+    //      "shell": "system"
+    // 2. A program:
+    //      "shell": {
+    //        "program": "sh"
+    //      }
+    // 3. A program with arguments:
+    //     "shell": {
+    //         "with_arguments": {
+    //           "program": "/bin/bash",
+    //           "args": ["--login"]
+    //         }
+    //     }
+    "shell": "system",
+    // Represents the tags for inline runnable indicators, or spawning multiple tasks at once.
+    "tags": []
+  }
+]

Makefile

@@ -2,6 +2,8 @@ run/new:
 	go run ./main.go
 run/latest:
 	go run ./main.go -n=false
+run/gemini:
+	go run ./main.go --provider=google
 serve:
 	go run ./main.go serve
 list/models:

agent/agent.go

@@ -88,6 +88,8 @@ func (a *Agent) Run(ctx context.Context) error {
 
 		for _, c := range agentMsg.Content {
 			switch c.Type {
+			// TODO: Switch case for text type should be here
+			// and we need to stream the response here, not inside the model integrations
 			case message.ToolUseType:
 				result := a.executeTool(c.OfToolUseBlock.ID, c.OfToolUseBlock.Name, c.OfToolUseBlock.Input)
 				toolResults = append(toolResults, result)
@@ -127,7 +129,7 @@ func (a *Agent) executeTool(id, name string, input json.RawMessage) message.Cont
 	if !found {
 		// TODO: Return proper error type
 		errorMsg := "tool not found"
-		return message.NewToolResultContentBlock(id, errorMsg, true)
+		return message.NewToolResultContentBlock(id, name, errorMsg, true)
 	}
 
 	fmt.Printf("\u001b[92mtool\u001b[0m: %s(%s)\n", name, input)
@@ -136,10 +138,10 @@ func (a *Agent) executeTool(id, name string, input json.RawMessage) message.Cont
 	response, err := toolDef.Function(input)
 
 	if err != nil {
-		return message.NewToolResultContentBlock(id, err.Error(), true)
+		return message.NewToolResultContentBlock(id, name, err.Error(), true)
 	}
 
-	return message.NewToolResultContentBlock(id, response, false)
+	return message.NewToolResultContentBlock(id, name, response, false)
 }
 
 func (a *Agent) saveConversation() error {

cmd/cmd.go

@@ -57,6 +57,11 @@ func ChatHandler(cmd *cobra.Command, args []string) error {
 		modelConfig.Model = string(defaultModel)
 	}
 
+	// Default number of max tokens
+	if modelConfig.MaxTokens == 0 {
+		modelConfig.MaxTokens = 8192
+	}
+
 	var convID string
 	if new {
 		convID = ""

cmd/interactive.go

@@ -15,7 +15,7 @@ import (
 )
 
 func interactive(ctx context.Context, convID string, modelConfig inference.ModelConfig, client *api.Client) error {
-	model, err := inference.Init(modelConfig)
+	model, err := inference.Init(ctx, modelConfig)
 	if err != nil {
 		log.Fatalf("Failed to initialize model: %s", err.Error())
 	}

devlogs/DevLogJuly2025.md

@@ -0,0 +1,5 @@
+- Default max token is now 8192. Larger context yay!
+
+- Gemini integration is here! With hacks and lots of TODOs of course, but I prioritize fast shipping :)
+
+- `grep_search` is really fun to build. Right now we need `ripgrep` as a dependency, but in the future I might try a native approach (less performant of course)

devlogs/DevLogJune2025.md

@@ -1,25 +1,15 @@
-[Delta vs snapshot streaming](https://docs.anthropic.com/en/docs/build-with-claude/streaming#delta-vs-snapshot-streaming)
+(This should have been done a month ago but I was lazy lol)
 
-Maybe a buffer is enoug? Get the stream, push the data from the stream to the buffer, then send the data from the buffer to a channel of CustomResponse? -> Go for this, dont overthink
+- I learn a lot about [Delta vs snapshot streaming](https://docs.anthropic.com/en/docs/build-with-claude/streaming#delta-vs-snapshot-streaming) when intergrating Claude and Gemini into the agent
 
-content_block_start events for tool use blocks
-content_block_delta events with accumulated partial JSON
+- At some point I was wondering about handling the streaming response from the model. Maybe a buffer is enough? Get the stream, push the data from the stream to the buffer, then send the data from the buffer to a channel of CustomResponse? I will try to keep it simple first. Maybe using an iterator is a better idea.
 
-`ContentBlock` as a unified interface for different content block types
+- `ContentBlockUnion` as a unified struct for different content block types. Stole this idea from "anthropic-sdk-go" although the implementation is kinda meh
 
-sqlite3 as a lightweight option to store conversations
+- I have always been thinking of implementing a server to handle all the CRUD-related stuff, just like what Ollama did with the models. And if there will be a server, there should be a database, and sqlite3 is perfect as a lightweight option to store conversations
 
-Two modes: Snapshot and streaming
+- A good piece of advice: Too many tools and the agent would get stuck and not know which one to use. A curated set of tools is the most important thing.
 
-Too many tools and the agent would stuck and not know which one to use. A curated set of tools is important
+- The tokens must flow. The agent should retry the operation instead of halting it
 
-The tokens must flow. The agent should retry the operation instead of halting it
-
-Next tools:
-
-Structural Search Interface (full‑text, regex, and language-aware structural code search)?
-Commit Diff Lookup
-
-Server to handle CRUD for conversations, lifecycle for server
-
-No need for store, that's for local model state
+- Ideas for the next tool(s): Commit Diff Lookup

go.mod

@@ -8,10 +8,19 @@ require (
 	github.com/invopop/jsonschema v0.13.0
 	github.com/mattn/go-sqlite3 v1.14.28
 	github.com/olekukonko/tablewriter v1.0.7
+	google.golang.org/genai v1.14.0
 )
 
 require (
+	cloud.google.com/go v0.116.0 // indirect
+	cloud.google.com/go/auth v0.9.3 // indirect
+	cloud.google.com/go/compute/metadata v0.5.0 // indirect
 	github.com/fatih/color v1.15.0 // indirect
+	github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da // indirect
+	github.com/google/go-cmp v0.6.0 // indirect
+	github.com/google/s2a-go v0.1.8 // indirect
+	github.com/googleapis/enterprise-certificate-proxy v0.3.4 // indirect
+	github.com/gorilla/websocket v1.5.3 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/mattn/go-colorable v0.1.13 // indirect
 	github.com/mattn/go-isatty v0.0.19 // indirect
@@ -20,7 +29,14 @@ require (
 	github.com/olekukonko/ll v0.0.8 // indirect
 	github.com/rivo/uniseg v0.2.0 // indirect
 	github.com/spf13/pflag v1.0.6 // indirect
-	golang.org/x/sys v0.22.0 // indirect
+	go.opencensus.io v0.24.0 // indirect
+	golang.org/x/crypto v0.27.0 // indirect
+	golang.org/x/net v0.29.0 // indirect
+	golang.org/x/sys v0.25.0 // indirect
+	golang.org/x/text v0.18.0 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20240903143218-8af14fe29dc1 // indirect
+	google.golang.org/grpc v1.66.2 // indirect
+	google.golang.org/protobuf v1.34.2 // indirect
 )
 
 require (