Add multimodal outputs to js_repl polling

git-stack-id: fjord/js_repl_seq---4hwa73o5rjfj-v
git-stack-title: Add multimodal outputs to js_repl polling

Author: fjord-oai

codex-rs/core/src/project_doc.rs

@@ -60,7 +60,7 @@ fn render_js_repl_instructions(config: &Config) -> Option<String> {
 
     if config.features.enabled(Feature::JsReplPolling) {
         section.push_str("- Polling mode is session-based: call `js_repl` with first-line pragma `// codex-js-repl: poll=true` to get `exec_id` and `session_id`; provide `session_id=<id>` in later `js_repl` calls to reuse that session state. Omit `session_id` to create a new polling session, and note that unknown `session_id` values fail. Use space-separated pragma arguments.\n");
-        section.push_str("- Use `js_repl_poll` with `exec_id` until `status` is `completed` or `error`; if a poll returns `status: running`, keep polling the same `exec_id` even if the logs or output already look complete. Poll responses also include `session_id`.\n");
+        section.push_str("- Use `js_repl_poll` with `exec_id` until `status` is `completed` or `error`; if a poll returns `status: running`, keep polling the same `exec_id` even if the logs or output already look complete. Poll responses also include `session_id`, and completed polls can also include nested multimodal tool output after the JSON status item.\n");
         section.push_str("- Use `js_repl_reset({\"session_id\":\"...\"})` to stop one polling session (including any running exec), or `js_repl_reset({})` to reset all js_repl kernels.\n");
         section.push_str("- `js_repl_poll` must not be called before a successful `js_repl` polling submission returns an `exec_id`.\n");
         section.push_str("- In polling mode, `timeout_ms` is not supported on `js_repl`; use `js_repl_poll` `yield_time_ms` to control poll wait duration. Omitted values, or values below `5000`, wait up to 5 seconds before returning if nothing new arrives.\n");
@@ -512,7 +512,7 @@ mod tests {
         let res = get_user_instructions(&cfg, None)
             .await
             .expect("js_repl instructions expected");
-        let expected = "## JavaScript REPL (Node)\n- Use `js_repl` for Node-backed JavaScript with top-level await in a persistent kernel.\n- `js_repl` is a freeform/custom tool. Direct `js_repl` calls must send raw JavaScript tool input (optionally with first-line `// codex-js-repl: timeout_ms=15000`). Do not wrap code in JSON (for example `{\"code\":\"...\"}`), quotes, or markdown code fences.\n- Helpers: `codex.tmpDir` and `codex.tool(name, args?)`.\n- `codex.tool` executes a normal tool call and resolves to the raw tool output object. Use it for shell and non-shell tools alike.\n- To share generated images with the model, write a file under `codex.tmpDir`, call `await codex.tool(\"view_image\", { path: \"/absolute/path\" })`, then delete the file.\n- Top-level bindings persist across cells. If you hit `SyntaxError: Identifier 'x' has already been declared`, reuse the binding, pick a new name, wrap in `{ ... }` for block scope, or reset the kernel with `js_repl_reset`.\n- Top-level static import declarations (for example `import x from \"pkg\"`) are currently unsupported in `js_repl`; use dynamic imports with `await import(\"pkg\")` instead.\n- Polling mode is session-based: call `js_repl` with first-line pragma `// codex-js-repl: poll=true` to get `exec_id` and `session_id`; provide `session_id=<id>` in later `js_repl` calls to reuse that session state. Omit `session_id` to create a new polling session, and note that unknown `session_id` values fail. Use space-separated pragma arguments.\n- Use `js_repl_poll` with `exec_id` until `status` is `completed` or `error`; if a poll returns `status: running`, keep polling the same `exec_id` even if the logs or output already look complete. Poll responses also include `session_id`.\n- Use `js_repl_reset({\"session_id\":\"...\"})` to stop one polling session (including any running exec), or `js_repl_reset({})` to reset all js_repl kernels.\n- `js_repl_poll` must not be called before a successful `js_repl` polling submission returns an `exec_id`.\n- In polling mode, `timeout_ms` is not supported on `js_repl`; use `js_repl_poll` `yield_time_ms` to control poll wait duration. Omitted values, or values below `5000`, wait up to 5 seconds before returning if nothing new arrives.\n- If `js_repl` rejects your payload format, resend raw JS with the pragma; do not retry with JSON, quoted strings, or markdown fences.\n- Avoid direct access to `process.stdout` / `process.stderr` / `process.stdin`; it can corrupt the JSON line protocol. Use `console.log` and `codex.tool(...)`.";
+        let expected = "## JavaScript REPL (Node)\n- Use `js_repl` for Node-backed JavaScript with top-level await in a persistent kernel.\n- `js_repl` is a freeform/custom tool. Direct `js_repl` calls must send raw JavaScript tool input (optionally with first-line `// codex-js-repl: timeout_ms=15000`). Do not wrap code in JSON (for example `{\"code\":\"...\"}`), quotes, or markdown code fences.\n- Helpers: `codex.tmpDir` and `codex.tool(name, args?)`.\n- `codex.tool` executes a normal tool call and resolves to the raw tool output object. Use it for shell and non-shell tools alike.\n- To share generated images with the model, write a file under `codex.tmpDir`, call `await codex.tool(\"view_image\", { path: \"/absolute/path\" })`, then delete the file.\n- Top-level bindings persist across cells. If you hit `SyntaxError: Identifier 'x' has already been declared`, reuse the binding, pick a new name, wrap in `{ ... }` for block scope, or reset the kernel with `js_repl_reset`.\n- Top-level static import declarations (for example `import x from \"pkg\"`) are currently unsupported in `js_repl`; use dynamic imports with `await import(\"pkg\")` instead.\n- Polling mode is session-based: call `js_repl` with first-line pragma `// codex-js-repl: poll=true` to get `exec_id` and `session_id`; provide `session_id=<id>` in later `js_repl` calls to reuse that session state. Omit `session_id` to create a new polling session, and note that unknown `session_id` values fail. Use space-separated pragma arguments.\n- Use `js_repl_poll` with `exec_id` until `status` is `completed` or `error`; if a poll returns `status: running`, keep polling the same `exec_id` even if the logs or output already look complete. Poll responses also include `session_id`, and completed polls can also include nested multimodal tool output after the JSON status item.\n- Use `js_repl_reset({\"session_id\":\"...\"})` to stop one polling session (including any running exec), or `js_repl_reset({})` to reset all js_repl kernels.\n- `js_repl_poll` must not be called before a successful `js_repl` polling submission returns an `exec_id`.\n- In polling mode, `timeout_ms` is not supported on `js_repl`; use `js_repl_poll` `yield_time_ms` to control poll wait duration. Omitted values, or values below `5000`, wait up to 5 seconds before returning if nothing new arrives.\n- If `js_repl` rejects your payload format, resend raw JS with the pragma; do not retry with JSON, quoted strings, or markdown fences.\n- Avoid direct access to `process.stdout` / `process.stderr` / `process.stdin`; it can corrupt the JSON line protocol. Use `console.log` and `codex.tool(...)`.";
         assert_eq!(res, expected);
     }
 

codex-rs/core/src/tools/handlers/js_repl.rs

@@ -367,9 +367,16 @@ fn format_poll_output(result: &JsExecPollResult) -> Result<JsReplPollOutput, Fun
         FunctionCallError::Fatal(format!("failed to serialize js_repl poll result: {err}"))
     })?;
 
-    Ok(JsReplPollOutput {
-        body: FunctionCallOutputBody::Text(content),
-    })
+    let body = if result.content_items.is_empty() {
+        FunctionCallOutputBody::Text(content)
+    } else {
+        let mut items = Vec::with_capacity(result.content_items.len() + 1);
+        items.push(FunctionCallOutputContentItem::InputText { text: content });
+        items.extend(result.content_items.clone());
+        FunctionCallOutputBody::ContentItems(items)
+    };
+
+    Ok(JsReplPollOutput { body })
 }
 
 fn parse_freeform_args(
@@ -532,6 +539,7 @@ mod tests {
     use crate::tools::js_repl::JS_REPL_TIMEOUT_ERROR_MESSAGE;
     use crate::tools::js_repl::JsExecPollResult;
     use codex_protocol::models::FunctionCallOutputBody;
+    use codex_protocol::models::FunctionCallOutputContentItem;
     use pretty_assertions::assert_eq;
     use serde_json::json;
 
@@ -661,6 +669,7 @@ mod tests {
             session_id: "session-1".to_string(),
             logs: vec!["line 1".to_string(), "line 2".to_string()],
             output: None,
+            content_items: Vec::new(),
             error: None,
             done: false,
         };
@@ -690,6 +699,7 @@ mod tests {
             session_id: "session-1".to_string(),
             logs: Vec::new(),
             output: Some(String::new()),
+            content_items: Vec::new(),
             error: None,
             done: true,
         };
@@ -712,6 +722,44 @@ mod tests {
         );
     }
 
+    #[test]
+    fn format_poll_output_serializes_multimodal_content_items() {
+        let result = JsExecPollResult {
+            exec_id: "exec-1".to_string(),
+            session_id: "session-1".to_string(),
+            logs: Vec::new(),
+            output: Some("stdout".to_string()),
+            content_items: vec![FunctionCallOutputContentItem::InputImage {
+                image_url: "data:image/png;base64,abc".to_string(),
+            }],
+            error: None,
+            done: true,
+        };
+        let output = format_poll_output(&result).expect("format poll output");
+        let FunctionCallOutputBody::ContentItems(items) = output.body else {
+            panic!("expected content item poll output");
+        };
+        assert_eq!(
+            items,
+            vec![
+                FunctionCallOutputContentItem::InputText {
+                    text: json!({
+                        "exec_id": "exec-1",
+                        "session_id": "session-1",
+                        "status": "completed",
+                        "logs": null,
+                        "output": "stdout",
+                        "error": null,
+                    })
+                    .to_string(),
+                },
+                FunctionCallOutputContentItem::InputImage {
+                    image_url: "data:image/png;base64,abc".to_string(),
+                },
+            ]
+        );
+    }
+
     #[test]
     fn js_repl_poll_args_reject_unknown_fields() {
         let err = serde_json::from_str::<super::JsReplPollArgs>(