@ai-sdk/openai 3.0.28 → 3.0.30

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # @ai-sdk/openai
 
+## 3.0.30
+
+### Patch Changes
+
+- ff12133: feat(provider/openai): support native skills and hosted shell
+
+## 3.0.29
+
+### Patch Changes
+
+- e2ee705: feat: differentiate text vs image input tokens
+
 ## 3.0.28
 
 ### Patch Changes

package/dist/index.d.mts

@@ -293,6 +293,21 @@ declare const openaiResponsesChunkSchema: _ai_sdk_provider_utils.LazySchema<{
         action: {
             commands: string[];
         };
+    } | {
+        type: "shell_call_output";
+        id: string;
+        call_id: string;
+        status: "completed" | "in_progress" | "incomplete";
+        output: {
+            stdout: string;
+            stderr: string;
+            outcome: {
+                type: "timeout";
+            } | {
+                type: "exit";
+                exit_code: number;
+            };
+        }[];
     };
 } | {
     type: "response.output_item.done";
@@ -439,6 +454,21 @@ declare const openaiResponsesChunkSchema: _ai_sdk_provider_utils.LazySchema<{
         action: {
             commands: string[];
         };
+    } | {
+        type: "shell_call_output";
+        id: string;
+        call_id: string;
+        status: "completed" | "in_progress" | "incomplete";
+        output: {
+            stdout: string;
+            stderr: string;
+            outcome: {
+                type: "timeout";
+            } | {
+                type: "exit";
+                exit_code: number;
+            };
+        }[];
     };
 } | {
     type: "response.function_call_arguments.delta";
@@ -700,7 +730,48 @@ declare const openaiTools: {
                 exitCode: number;
             };
         }>;
-    }, {}>;
+    }, {
+        environment?: {
+            type: "containerAuto";
+            fileIds?: string[];
+            memoryLimit?: "1g" | "4g" | "16g" | "64g";
+            networkPolicy?: {
+                type: "disabled";
+            } | {
+                type: "allowlist";
+                allowedDomains: string[];
+                domainSecrets?: Array<{
+                    domain: string;
+                    name: string;
+                    value: string;
+                }>;
+            };
+            skills?: Array<{
+                type: "skillReference";
+                skillId: string;
+                version?: string;
+            } | {
+                type: "inline";
+                name: string;
+                description: string;
+                source: {
+                    type: "base64";
+                    mediaType: "application/zip";
+                    data: string;
+                };
+            }>;
+        } | {
+            type: "containerReference";
+            containerId: string;
+        } | {
+            type?: "local";
+            skills?: Array<{
+                name: string;
+                description: string;
+                path: string;
+            }>;
+        };
+    }>;
     /**
      * Web search allows models to access up-to-date information from the internet
      * and provide answers with sourced citations.

package/dist/index.d.ts

@@ -293,6 +293,21 @@ declare const openaiResponsesChunkSchema: _ai_sdk_provider_utils.LazySchema<{
         action: {
             commands: string[];
         };
+    } | {
+        type: "shell_call_output";
+        id: string;
+        call_id: string;
+        status: "completed" | "in_progress" | "incomplete";
+        output: {
+            stdout: string;
+            stderr: string;
+            outcome: {
+                type: "timeout";
+            } | {
+                type: "exit";
+                exit_code: number;
+            };
+        }[];
     };
 } | {
     type: "response.output_item.done";
@@ -439,6 +454,21 @@ declare const openaiResponsesChunkSchema: _ai_sdk_provider_utils.LazySchema<{
         action: {
             commands: string[];
         };
+    } | {
+        type: "shell_call_output";
+        id: string;
+        call_id: string;
+        status: "completed" | "in_progress" | "incomplete";
+        output: {
+            stdout: string;
+            stderr: string;
+            outcome: {
+                type: "timeout";
+            } | {
+                type: "exit";
+                exit_code: number;
+            };
+        }[];
     };
 } | {
     type: "response.function_call_arguments.delta";
@@ -700,7 +730,48 @@ declare const openaiTools: {
                 exitCode: number;
             };
         }>;
-    }, {}>;
+    }, {
+        environment?: {
+            type: "containerAuto";
+            fileIds?: string[];
+            memoryLimit?: "1g" | "4g" | "16g" | "64g";
+            networkPolicy?: {
+                type: "disabled";
+            } | {
+                type: "allowlist";
+                allowedDomains: string[];
+                domainSecrets?: Array<{
+                    domain: string;
+                    name: string;
+                    value: string;
+                }>;
+            };
+            skills?: Array<{
+                type: "skillReference";
+                skillId: string;
+                version?: string;
+            } | {
+                type: "inline";
+                name: string;
+                description: string;
+                source: {
+                    type: "base64";
+                    mediaType: "application/zip";
+                    data: string;
+                };
+            }>;
+        } | {
+            type: "containerReference";
+            containerId: string;
+        } | {
+            type?: "local";
+            skills?: Array<{
+                name: string;
+                description: string;
+                path: string;
+            }>;
+        };
+    }>;
     /**
      * Web search allows models to access up-to-date information from the internet
      * and provide answers with sourced citations.

package/dist/index.js

@@ -1852,15 +1852,20 @@ var OpenAIImageModel = class {
         },
         providerMetadata: {
           openai: {
-            images: response2.data.map((item) => {
-              var _a2, _b2, _c2, _d2, _e2;
+            images: response2.data.map((item, index) => {
+              var _a2, _b2, _c2, _d2, _e2, _f2;
               return {
                 ...item.revised_prompt ? { revisedPrompt: item.revised_prompt } : {},
                 created: (_a2 = response2.created) != null ? _a2 : void 0,
                 size: (_b2 = response2.size) != null ? _b2 : void 0,
                 quality: (_c2 = response2.quality) != null ? _c2 : void 0,
                 background: (_d2 = response2.background) != null ? _d2 : void 0,
-                outputFormat: (_e2 = response2.output_format) != null ? _e2 : void 0
+                outputFormat: (_e2 = response2.output_format) != null ? _e2 : void 0,
+                ...distributeTokenDetails(
+                  (_f2 = response2.usage) == null ? void 0 : _f2.input_tokens_details,
+                  index,
+                  response2.data.length
+                )
               };
             })
           }
@@ -1903,15 +1908,20 @@ var OpenAIImageModel = class {
       },
       providerMetadata: {
         openai: {
-          images: response.data.map((item) => {
-            var _a2, _b2, _c2, _d2, _e2;
+          images: response.data.map((item, index) => {
+            var _a2, _b2, _c2, _d2, _e2, _f2;
             return {
               ...item.revised_prompt ? { revisedPrompt: item.revised_prompt } : {},
               created: (_a2 = response.created) != null ? _a2 : void 0,
               size: (_b2 = response.size) != null ? _b2 : void 0,
               quality: (_c2 = response.quality) != null ? _c2 : void 0,
               background: (_d2 = response.background) != null ? _d2 : void 0,
-              outputFormat: (_e2 = response.output_format) != null ? _e2 : void 0
+              outputFormat: (_e2 = response.output_format) != null ? _e2 : void 0,
+              ...distributeTokenDetails(
+                (_f2 = response.usage) == null ? void 0 : _f2.input_tokens_details,
+                index,
+                response.data.length
+              )
             };
           })
         }
@@ -1919,6 +1929,23 @@ var OpenAIImageModel = class {
     };
   }
 };
+function distributeTokenDetails(details, index, total) {
+  if (details == null) {
+    return {};
+  }
+  const result = {};
+  if (details.image_tokens != null) {
+    const base = Math.floor(details.image_tokens / total);
+    const remainder = details.image_tokens - base * (total - 1);
+    result.imageTokens = index === total - 1 ? remainder : base;
+  }
+  if (details.text_tokens != null) {
+    const base = Math.floor(details.text_tokens / total);
+    const remainder = details.text_tokens - base * (total - 1);
+    result.textTokens = index === total - 1 ? remainder : base;
+  }
+  return result;
+}
 async function fileToBlob(file) {
   if (!file) return void 0;
   if (file.type === "url") {
@@ -2154,6 +2181,67 @@ var shellOutputSchema = (0, import_provider_utils19.lazySchema)(
     })
   )
 );
+var shellSkillsSchema = import_v414.z.array(
+  import_v414.z.discriminatedUnion("type", [
+    import_v414.z.object({
+      type: import_v414.z.literal("skillReference"),
+      skillId: import_v414.z.string(),
+      version: import_v414.z.string().optional()
+    }),
+    import_v414.z.object({
+      type: import_v414.z.literal("inline"),
+      name: import_v414.z.string(),
+      description: import_v414.z.string(),
+      source: import_v414.z.object({
+        type: import_v414.z.literal("base64"),
+        mediaType: import_v414.z.literal("application/zip"),
+        data: import_v414.z.string()
+      })
+    })
+  ])
+).optional();
+var shellArgsSchema = (0, import_provider_utils19.lazySchema)(
+  () => (0, import_provider_utils19.zodSchema)(
+    import_v414.z.object({
+      environment: import_v414.z.union([
+        import_v414.z.object({
+          type: import_v414.z.literal("containerAuto"),
+          fileIds: import_v414.z.array(import_v414.z.string()).optional(),
+          memoryLimit: import_v414.z.enum(["1g", "4g", "16g", "64g"]).optional(),
+          networkPolicy: import_v414.z.discriminatedUnion("type", [
+            import_v414.z.object({ type: import_v414.z.literal("disabled") }),
+            import_v414.z.object({
+              type: import_v414.z.literal("allowlist"),
+              allowedDomains: import_v414.z.array(import_v414.z.string()),
+              domainSecrets: import_v414.z.array(
+                import_v414.z.object({
+                  domain: import_v414.z.string(),
+                  name: import_v414.z.string(),
+                  value: import_v414.z.string()
+                })
+              ).optional()
+            })
+          ]).optional(),
+          skills: shellSkillsSchema
+        }),
+        import_v414.z.object({
+          type: import_v414.z.literal("containerReference"),
+          containerId: import_v414.z.string()
+        }),
+        import_v414.z.object({
+          type: import_v414.z.literal("local").optional(),
+          skills: import_v414.z.array(
+            import_v414.z.object({
+              name: import_v414.z.string(),
+              description: import_v414.z.string(),
+              path: import_v414.z.string()
+            })
+          ).optional()
+        })
+      ]).optional()
+    })
+  )
+);
 var shell = (0, import_provider_utils19.createProviderToolFactoryWithOutputSchema)({
   id: "openai.shell",
   inputSchema: shellInputSchema,
@@ -2668,8 +2756,32 @@ async function convertToOpenAIResponsesInput({
               if (hasConversation) {
                 break;
               }
+              const resolvedResultToolName = toolNameMapping.toProviderToolName(
+                part.toolName
+              );
+              if (hasShellTool && resolvedResultToolName === "shell") {
+                if (part.output.type === "json") {
+                  const parsedOutput = await (0, import_provider_utils23.validateTypes)({
+                    value: part.output.value,
+                    schema: shellOutputSchema
+                  });
+                  input.push({
+                    type: "shell_call_output",
+                    call_id: part.toolCallId,
+                    output: parsedOutput.output.map((item) => ({
+                      stdout: item.stdout,
+                      stderr: item.stderr,
+                      outcome: item.outcome.type === "timeout" ? { type: "timeout" } : {
+                        type: "exit",
+                        exit_code: item.outcome.exitCode
+                      }
+                    }))
+                  });
+                }
+                break;
+              }
               if (store) {
-                const itemId = (_j = (_i = (_h = part.providerMetadata) == null ? void 0 : _h[providerOptionsName]) == null ? void 0 : _i.itemId) != null ? _j : part.toolCallId;
+                const itemId = (_j = (_i = (_h = part.providerOptions) == null ? void 0 : _h[providerOptionsName]) == null ? void 0 : _i.itemId) != null ? _j : part.toolCallId;
                 input.push({ type: "item_reference", id: itemId });
               } else {
                 warnings.push({
@@ -3044,6 +3156,25 @@ var openaiResponsesChunkSchema = (0, import_provider_utils24.lazySchema)(
             action: import_v419.z.object({
               commands: import_v419.z.array(import_v419.z.string())
             })
+          }),
+          import_v419.z.object({
+            type: import_v419.z.literal("shell_call_output"),
+            id: import_v419.z.string(),
+            call_id: import_v419.z.string(),
+            status: import_v419.z.enum(["in_progress", "completed", "incomplete"]),
+            output: import_v419.z.array(
+              import_v419.z.object({
+                stdout: import_v419.z.string(),
+                stderr: import_v419.z.string(),
+                outcome: import_v419.z.discriminatedUnion("type", [
+                  import_v419.z.object({ type: import_v419.z.literal("timeout") }),
+                  import_v419.z.object({
+                    type: import_v419.z.literal("exit"),
+                    exit_code: import_v419.z.number()
+                  })
+                ])
+              })
+            )
           })
         ])
       }),
@@ -3223,6 +3354,25 @@ var openaiResponsesChunkSchema = (0, import_provider_utils24.lazySchema)(
             action: import_v419.z.object({
               commands: import_v419.z.array(import_v419.z.string())
             })
+          }),
+          import_v419.z.object({
+            type: import_v419.z.literal("shell_call_output"),
+            id: import_v419.z.string(),
+            call_id: import_v419.z.string(),
+            status: import_v419.z.enum(["in_progress", "completed", "incomplete"]),
+            output: import_v419.z.array(
+              import_v419.z.object({
+                stdout: import_v419.z.string(),
+                stderr: import_v419.z.string(),
+                outcome: import_v419.z.discriminatedUnion("type", [
+                  import_v419.z.object({ type: import_v419.z.literal("timeout") }),
+                  import_v419.z.object({
+                    type: import_v419.z.literal("exit"),
+                    exit_code: import_v419.z.number()
+                  })
+                ])
+              })
+            )
           })
         ])
       }),
@@ -3568,6 +3718,25 @@ var openaiResponsesResponseSchema = (0, import_provider_utils24.lazySchema)(
             action: import_v419.z.object({
               commands: import_v419.z.array(import_v419.z.string())
             })
+          }),
+          import_v419.z.object({
+            type: import_v419.z.literal("shell_call_output"),
+            id: import_v419.z.string(),
+            call_id: import_v419.z.string(),
+            status: import_v419.z.enum(["in_progress", "completed", "incomplete"]),
+            output: import_v419.z.array(
+              import_v419.z.object({
+                stdout: import_v419.z.string(),
+                stderr: import_v419.z.string(),
+                outcome: import_v419.z.discriminatedUnion("type", [
+                  import_v419.z.object({ type: import_v419.z.literal("timeout") }),
+                  import_v419.z.object({
+                    type: import_v419.z.literal("exit"),
+                    exit_code: import_v419.z.number()
+                  })
+                ])
+              })
+            )
           })
         ])
       ).optional(),
@@ -3861,8 +4030,15 @@ async function prepareResponsesTools({
             break;
           }
           case "openai.shell": {
+            const args = await (0, import_provider_utils26.validateTypes)({
+              value: tool.args,
+              schema: shellArgsSchema
+            });
             openaiTools2.push({
-              type: "shell"
+              type: "shell",
+              ...args.environment && {
+                environment: mapShellEnvironment(args.environment)
+              }
             });
             break;
           }
@@ -3992,6 +4168,52 @@ async function prepareResponsesTools({
     }
   }
 }
+function mapShellEnvironment(environment) {
+  if (environment.type === "containerReference") {
+    const env2 = environment;
+    return {
+      type: "container_reference",
+      container_id: env2.containerId
+    };
+  }
+  if (environment.type === "containerAuto") {
+    const env2 = environment;
+    return {
+      type: "container_auto",
+      file_ids: env2.fileIds,
+      memory_limit: env2.memoryLimit,
+      network_policy: env2.networkPolicy == null ? void 0 : env2.networkPolicy.type === "disabled" ? { type: "disabled" } : {
+        type: "allowlist",
+        allowed_domains: env2.networkPolicy.allowedDomains,
+        domain_secrets: env2.networkPolicy.domainSecrets
+      },
+      skills: mapShellSkills(env2.skills)
+    };
+  }
+  const env = environment;
+  return {
+    type: "local",
+    skills: env.skills
+  };
+}
+function mapShellSkills(skills) {
+  return skills == null ? void 0 : skills.map(
+    (skill) => skill.type === "skillReference" ? {
+      type: "skill_reference",
+      skill_id: skill.skillId,
+      version: skill.version
+    } : {
+      type: "inline",
+      name: skill.name,
+      description: skill.description,
+      source: {
+        type: "base64",
+        media_type: skill.source.mediaType,
+        data: skill.source.data
+      }
+    }
+  );
+}
 
 // src/responses/openai-responses-language-model.ts
 function extractApprovalRequestIdToToolCallIdMapping(prompt) {
@@ -4037,7 +4259,7 @@ var OpenAIResponsesLanguageModel = class {
     toolChoice,
     responseFormat
   }) {
-    var _a, _b, _c, _d, _e, _f;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i;
     const warnings = [];
     const modelCapabilities = getOpenAILanguageModelCapabilities(this.modelId);
     if (topK != null) {
@@ -4241,6 +4463,10 @@ var OpenAIResponsesLanguageModel = class {
       tools,
       toolChoice
     });
+    const shellToolEnvType = (_i = (_h = (_g = tools == null ? void 0 : tools.find(
+      (tool) => tool.type === "provider" && tool.id === "openai.shell"
+    )) == null ? void 0 : _g.args) == null ? void 0 : _h.environment) == null ? void 0 : _i.type;
+    const isShellProviderExecuted = shellToolEnvType === "containerAuto" || shellToolEnvType === "containerReference";
     return {
       webSearchToolName,
       args: {
@@ -4251,7 +4477,8 @@ var OpenAIResponsesLanguageModel = class {
       warnings: [...warnings, ...toolWarnings],
       store,
       toolNameMapping,
-      providerOptionsName
+      providerOptionsName,
+      isShellProviderExecuted
     };
   }
   async doGenerate(options) {
@@ -4261,7 +4488,8 @@ var OpenAIResponsesLanguageModel = class {
       warnings,
       webSearchToolName,
       toolNameMapping,
-      providerOptionsName
+      providerOptionsName,
+      isShellProviderExecuted
     } = await this.getArgs(options);
     const url = this.config.url({
       path: "/responses",
@@ -4361,6 +4589,7 @@ var OpenAIResponsesLanguageModel = class {
                 commands: part.action.commands
               }
             }),
+            ...isShellProviderExecuted && { providerExecuted: true },
             providerMetadata: {
               [providerOptionsName]: {
                 itemId: part.id
@@ -4369,6 +4598,24 @@ var OpenAIResponsesLanguageModel = class {
           });
           break;
         }
+        case "shell_call_output": {
+          content.push({
+            type: "tool-result",
+            toolCallId: part.call_id,
+            toolName: toolNameMapping.toCustomToolName("shell"),
+            result: {
+              output: part.output.map((item) => ({
+                stdout: item.stdout,
+                stderr: item.stderr,
+                outcome: item.outcome.type === "exit" ? {
+                  type: "exit",
+                  exitCode: item.outcome.exit_code
+                } : { type: "timeout" }
+              }))
+            }
+          });
+          break;
+        }
         case "message": {
           for (const contentPart of part.content) {
             if (((_c = (_b = options.providerOptions) == null ? void 0 : _b[providerOptionsName]) == null ? void 0 : _c.logprobs) && contentPart.logprobs) {
@@ -4658,7 +4905,8 @@ var OpenAIResponsesLanguageModel = class {
       webSearchToolName,
       toolNameMapping,
       store,
-      providerOptionsName
+      providerOptionsName,
+      isShellProviderExecuted
     } = await this.getArgs(options);
     const { responseHeaders, value: response } = await (0, import_provider_utils27.postJsonToApi)({
       url: this.config.url({
@@ -4837,6 +5085,7 @@ var OpenAIResponsesLanguageModel = class {
                   toolName: toolNameMapping.toCustomToolName("shell"),
                   toolCallId: value.item.call_id
                 };
+              } else if (value.item.type === "shell_call_output") {
               } else if (value.item.type === "message") {
                 ongoingAnnotations.splice(0, ongoingAnnotations.length);
                 controller.enqueue({
@@ -5090,10 +5339,31 @@ var OpenAIResponsesLanguageModel = class {
                       commands: value.item.action.commands
                     }
                   }),
+                  ...isShellProviderExecuted && {
+                    providerExecuted: true
+                  },
                   providerMetadata: {
                     [providerOptionsName]: { itemId: value.item.id }
                   }
                 });
+              } else if (value.item.type === "shell_call_output") {
+                controller.enqueue({
+                  type: "tool-result",
+                  toolCallId: value.item.call_id,
+                  toolName: toolNameMapping.toCustomToolName("shell"),
+                  result: {
+                    output: value.item.output.map(
+                      (item) => ({
+                        stdout: item.stdout,
+                        stderr: item.stderr,
+                        outcome: item.outcome.type === "exit" ? {
+                          type: "exit",
+                          exitCode: item.outcome.exit_code
+                        } : { type: "timeout" }
+                      })
+                    )
+                  }
+                });
               } else if (value.item.type === "reasoning") {
                 const activeReasoningPart = activeReasoning[value.item.id];
                 const summaryPartIndices = Object.entries(
@@ -5786,7 +6056,7 @@ var OpenAITranscriptionModel = class {
 };
 
 // src/version.ts
-var VERSION = true ? "3.0.28" : "0.0.0-test";
+var VERSION = true ? "3.0.30" : "0.0.0-test";
 
 // src/openai-provider.ts
 function createOpenAI(options = {}) {