image-gen-core (#13290)

Core tool-calling for image-gen, handles requesting and receiving logic
for images using response API

codex-rs/core/src/tools/spec.rs

@@ -23,6 +23,7 @@ use codex_protocol::dynamic_tools::DynamicToolSpec;
 use codex_protocol::models::VIEW_IMAGE_TOOL_NAME;
 use codex_protocol::openai_models::ApplyPatchToolType;
 use codex_protocol::openai_models::ConfigShellToolType;
+use codex_protocol::openai_models::InputModality;
 use codex_protocol::openai_models::ModelInfo;
 use codex_protocol::protocol::SessionSource;
 use codex_protocol::protocol::SubAgentSource;
@@ -35,7 +36,6 @@ use std::collections::HashMap;
 
 const SEARCH_TOOL_BM25_DESCRIPTION_TEMPLATE: &str =
     include_str!("../../templates/search_tool/tool_description.md");
-
 #[derive(Debug, Clone, Copy, Eq, PartialEq)]
 pub enum ShellCommandBackendConfig {
     Classic,
@@ -49,6 +49,7 @@ pub(crate) struct ToolsConfig {
     pub allow_login_shell: bool,
     pub apply_patch_tool_type: Option<ApplyPatchToolType>,
     pub web_search_mode: Option<WebSearchMode>,
+    pub image_gen_tool: bool,
     pub agent_roles: BTreeMap<String, AgentRoleConfig>,
     pub search_tool: bool,
     pub request_permission_enabled: bool,
@@ -86,6 +87,8 @@ impl ToolsConfig {
             features.enabled(Feature::DefaultModeRequestUserInput);
         let include_search_tool = features.enabled(Feature::Apps);
         let include_artifact_tools = features.enabled(Feature::Artifact);
+        let include_image_gen_tool =
+            features.enabled(Feature::ImageGeneration) && supports_image_generation(model_info);
         let include_agent_jobs = include_collab_tools && features.enabled(Feature::Sqlite);
         let request_permission_enabled = features.enabled(Feature::RequestPermissions);
         let shell_command_backend =
@@ -135,6 +138,7 @@ impl ToolsConfig {
             allow_login_shell: true,
             apply_patch_tool_type,
             web_search_mode: *web_search_mode,
+            image_gen_tool: include_image_gen_tool,
             agent_roles: BTreeMap::new(),
             search_tool: include_search_tool,
             request_permission_enabled,
@@ -160,6 +164,10 @@ impl ToolsConfig {
     }
 }
 
+fn supports_image_generation(model_info: &ModelInfo) -> bool {
+    model_info.input_modalities.contains(&InputModality::Image)
+}
+
 /// Generic JSON‑Schema subset needed for our tool definitions
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
 #[serde(tag = "type", rename_all = "lowercase")]
@@ -1917,6 +1925,10 @@ pub(crate) fn build_specs(
         Some(WebSearchMode::Disabled) | None => {}
     }
 
+    if config.image_gen_tool {
+        builder.push_spec(ToolSpec::ImageGeneration {});
+    }
+
     builder.push_spec_with_parallel_support(create_view_image_tool(), true);
     builder.register_handler("view_image", view_image_handler);
 
@@ -1995,6 +2007,7 @@ mod tests {
     use crate::models_manager::manager::ModelsManager;
     use crate::models_manager::model_info::with_config_overrides;
     use crate::tools::registry::ConfiguredToolSpec;
+    use codex_protocol::openai_models::InputModality;
     use codex_protocol::openai_models::ModelInfo;
     use codex_protocol::openai_models::ModelsResponse;
     use pretty_assertions::assert_eq;
@@ -2047,6 +2060,7 @@ mod tests {
         match tool {
             ToolSpec::Function(ResponsesApiTool { name, .. }) => name,
             ToolSpec::LocalShell {} => "local_shell",
+            ToolSpec::ImageGeneration {} => "image_generation",
             ToolSpec::WebSearch { .. } => "web_search",
             ToolSpec::Freeform(FreeformTool { name, .. }) => name,
         }
@@ -2125,7 +2139,10 @@ mod tests {
             ToolSpec::Function(ResponsesApiTool { parameters, .. }) => {
                 strip_descriptions_schema(parameters);
             }
-            ToolSpec::Freeform(_) | ToolSpec::LocalShell {} | ToolSpec::WebSearch { .. } => {}
+            ToolSpec::Freeform(_)
+            | ToolSpec::LocalShell {}
+            | ToolSpec::ImageGeneration {}
+            | ToolSpec::WebSearch { .. } => {}
         }
     }
 
@@ -2374,6 +2391,56 @@ mod tests {
         assert_contains_tool_names(&tools, &["js_repl", "js_repl_reset"]);
     }
 
+    #[test]
+    fn image_generation_tools_require_feature_and_supported_model() {
+        let config = test_config();
+        let mut supported_model_info =
+            ModelsManager::construct_model_info_offline_for_tests("gpt-5.2", &config);
+        supported_model_info.slug = "custom/gpt-5.2-variant".to_string();
+        let mut unsupported_model_info = supported_model_info.clone();
+        unsupported_model_info.input_modalities = vec![InputModality::Text];
+        let default_features = Features::with_defaults();
+        let mut image_generation_features = default_features.clone();
+        image_generation_features.enable(Feature::ImageGeneration);
+
+        let default_tools_config = ToolsConfig::new(&ToolsConfigParams {
+            model_info: &supported_model_info,
+            features: &default_features,
+            web_search_mode: Some(WebSearchMode::Cached),
+            session_source: SessionSource::Cli,
+        });
+        let (default_tools, _) = build_specs(&default_tools_config, None, None, &[]).build();
+        assert!(
+            !default_tools
+                .iter()
+                .any(|tool| tool.spec.name() == "image_generation"),
+            "image_generation should be disabled by default"
+        );
+
+        let supported_tools_config = ToolsConfig::new(&ToolsConfigParams {
+            model_info: &supported_model_info,
+            features: &image_generation_features,
+            web_search_mode: Some(WebSearchMode::Cached),
+            session_source: SessionSource::Cli,
+        });
+        let (supported_tools, _) = build_specs(&supported_tools_config, None, None, &[]).build();
+        assert_contains_tool_names(&supported_tools, &["image_generation"]);
+
+        let tools_config = ToolsConfig::new(&ToolsConfigParams {
+            model_info: &unsupported_model_info,
+            features: &image_generation_features,
+            web_search_mode: Some(WebSearchMode::Cached),
+            session_source: SessionSource::Cli,
+        });
+        let (tools, _) = build_specs(&tools_config, None, None, &[]).build();
+        assert!(
+            !tools
+                .iter()
+                .any(|tool| tool.spec.name() == "image_generation"),
+            "image_generation should be disabled for unsupported models"
+        );
+    }
+
     #[test]
     fn js_repl_freeform_grammar_blocks_common_non_js_prefixes() {
         let ToolSpec::Freeform(FreeformTool { format, .. }) = create_js_repl_tool() else {