@ai-sdk/harness-codex 1.0.165 → 1.0.167

package/CHANGELOG.md

@@ -1,5 +1,27 @@
 # @ai-sdk/harness-codex
 
+## 1.0.167
+
+### Patch Changes
+
+- 39aae54: fix(harness): ensure harness adapters always resolve the model used
+- Updated dependencies [e64ed82]
+  - @ai-sdk/provider-utils@5.0.69
+  - @ai-sdk/harness@1.0.165
+
+## 1.0.166
+
+### Patch Changes
+
+- a9837eb: feat(harness): provide shared utils for bridge-backed adapters to manage web socket connection and bridge handshake
+- e729d58: chore(docs): update `README.md` for consistent format and to avoid duplication
+- Updated dependencies [5475347]
+- Updated dependencies [a9837eb]
+- Updated dependencies [e729d58]
+  - @ai-sdk/provider@4.0.34
+  - @ai-sdk/harness@1.0.164
+  - @ai-sdk/provider-utils@5.0.68
+
 ## 1.0.165
 
 ### Patch Changes

package/README.md

@@ -2,14 +2,14 @@
 
 `HarnessV1` adapter backed by the [Codex CLI](https://www.npmjs.com/package/@openai/codex). The adapter runs Codex app-server inside a sandbox and communicates with it over JSON-RPC. A bridge process connects app-server to the host over a WebSocket on a sandbox-proxied loopback port.
 
+-> **Note**: This package is experimental.
+
 ## Setup
 
 ```bash
-npm i @ai-sdk/harness-codex @ai-sdk/harness @ai-sdk/sandbox-vercel
+npm i ai @ai-sdk/harness-codex @ai-sdk/harness @ai-sdk/sandbox-vercel zod
 ```
 
-The bridge installs the Codex CLI inside the sandbox the first time the session starts.
-
 ## Usage
 
 ```ts
@@ -39,43 +39,4 @@ const agent = new HarnessAgent({
 });
 ```
 
-`codexConfig` accepts additional native Codex configuration. Values pass
-through as provided, so use the snake_case keys from Codex's `config.toml`
-reference. The adapter's managed values take precedence over conflicting
-entries.
-
-> Codex does not auto-discover a skills directory the way the `claude` CLI
-> does, so when you supply `skills: [...]` on the factory the adapter
-> injects every skill inline into the user prompt on each turn. Use fewer,
-> larger skills rather than many tiny ones.
-
-```ts
-const agent = new HarnessAgent({
-  harness: createCodex({
-    skills: [
-      { name: 'haiku-mode', description: 'Answer in haikus.', content: '...' },
-    ],
-  }),
-});
-
-const sandboxSession = await createVercelNetworkSandboxSession({
-  runtime: 'node24',
-  ports: [4000],
-  template: await agent.getSandboxTemplate(),
-});
-const session = await agent.createSession({ sandboxSession });
-
-try {
-  const result = await agent.generate({
-    session,
-    prompt: 'List the files in this workspace and describe their purpose.',
-  });
-  console.log(result.text);
-} finally {
-  await session.destroy();
-  await sandboxSession.destroy();
-}
-```
-
-The adapter needs a sandbox session with an exposed port. The caller ends the
-harness session and sandbox separately.
+See the [Codex Harness documentation](https://ai-sdk.dev/providers/ai-sdk-harnesses/codex) for more information.

package/dist/bridge/index.mjs

@@ -1342,12 +1342,14 @@ const CODEX_CLI_PATH = fileURLToPath(new URL("./node_modules/@openai/codex/bin/c
 function createCodexAppServerRuntime() {
 	let client;
 	let loadedThreadId;
+	let loadedThreadModel;
 	let loadedConfig;
 	let activeTurn;
 	const close = async () => {
 		const previous = client;
 		client = void 0;
 		loadedThreadId = void 0;
+		loadedThreadModel = void 0;
 		loadedConfig = void 0;
 		await previous?.close();
 	};
@@ -1505,6 +1507,7 @@ function createCodexAppServerRuntime() {
 					params: { threadId: loadedThreadId }
 				}) });
 				loadedThreadId = void 0;
+				loadedThreadModel = void 0;
 			}
 			if (trustedHook != null) assertTrustedApplyPatchHook({
 				response: await raceWithProcess({ operation: runningClient.request({
@@ -1547,9 +1550,18 @@ function createCodexAppServerRuntime() {
 					method: threadMethod
 				});
 				loadedThreadId = currentTurn.threadId;
+				loadedThreadModel = {
+					requestedModel: codexModel,
+					modelId: readOptionalString(asRecord(threadResponse)?.model)
+				};
 			}
 			handler.announceThread(currentTurn.threadId);
-			emit({ type: "stream-start" });
+			const threadModel = loadedThreadModel;
+			const streamStartModelId = threadModel != null && threadModel.requestedModel === codexModel && threadModel.modelId != null ? threadModel.modelId : codexModel;
+			emit({
+				type: "stream-start",
+				...streamStartModelId ? { modelId: streamStartModelId } : {}
+			});
 			currentTurn.turnId = readNestedString({
 				value: await raceWithProcess({ operation: runningClient.request({
 					method: "turn/start",
@@ -1746,6 +1758,9 @@ function assertSuccessfulTurn(result) {
 function asRecord(value) {
 	return value != null && typeof value === "object" && !Array.isArray(value) ? value : void 0;
 }
+function readOptionalString(value) {
+	return typeof value === "string" && value.length > 0 ? value : void 0;
+}
 //#endregion
 //#region src/bridge/codex-step-tracker.ts
 function createCodexStepTracker(input) {