@mindstudio-ai/remy 0.1.318 → 0.1.320

package/dist/automatedActions/reviewExistingProject.md

@@ -0,0 +1,7 @@
+---
+trigger: reviewExistingProject
+---
+
+This is an automated message triggered by the user choosing "Bring an existing project" on the Remy start screen and uploading what they have. They can't see this message. The uploads are attached and already saved under `src/.user-uploads/` (paths are in the attachment header; a folder upload arrives as one zip). Repository URL, blank if none: {{repoUrl}}
+
+Let the user know you're going to review their project, and then hand the upload paths (and the URL, if any) to `reviewExistingProject` first, before saying anything substantive, and do not open the archive yourself. Then run intake as usual on top of what comes back: tell the user in a few lines what you found and what you think they were building, and put the review's questions to them as a form. The prior attempt is evidence of intent, not a spec. Oftentimes, the user is uploading something that never worked or was half-finished (hence why they're here!). What actually gets built in this session is decided with the user, not inherited from the upload. Treat the results of the review as a *communication shortcut* to more quickly align with the user, not as a guide for "importing" their project!

package/dist/headless.js

@@ -441,6 +441,13 @@ var MODEL_SURFACES = {
     modelType: "text",
     userPickable: true
   },
+  reviewExistingProject: {
+    default: "claude-5-sonnet",
+    label: "Existing Project Review",
+    description: "Reviews a project you bring from another tool and reports what is worth carrying forward.",
+    modelType: "text",
+    userPickable: true
+  },
   copyEditor: {
     default: "claude-5-sonnet",
     label: "Copy Agent",
@@ -512,6 +519,7 @@ var TEXT_MODELS = {
   "gpt-5.6-sol": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
   "gpt-5.6-terra": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
   "gpt-5.6-luna": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
+  "gpt-6-astra": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
   // Google ~1M-context; only 3.1-pro is tiered (higher rates above 200K).
   "gemini-3-pro": { forceCompactAt: 85e4 },
   "gemini-3.1-pro": { forceCompactAt: 85e4, suggestCompactAt: 18e4 },
@@ -1419,6 +1427,9 @@ async function listRecursive(dir) {
   for (const entry of entries) {
     const fullPath = path5.join(dir, entry.name);
     if (entry.isDirectory()) {
+      if (entry.name.startsWith(".")) {
+        continue;
+      }
       results.push(`${fullPath}/`);
       results.push(...await listRecursive(fullPath));
     } else {
@@ -2645,8 +2656,8 @@ function formatSize(bytes) {
 }
 async function formatFile(dirPath, name, indent) {
   try {
-    const stat4 = await fs15.stat(path8.join(dirPath, name));
-    return `${indent}${name}${" ".repeat(Math.max(1, 30 - indent.length - name.length))}${formatSize(stat4.size)}`;
+    const stat6 = await fs15.stat(path8.join(dirPath, name));
+    return `${indent}${name}${" ".repeat(Math.max(1, 30 - indent.length - name.length))}${formatSize(stat6.size)}`;
   } catch {
     return `${indent}${name}`;
   }
@@ -3436,6 +3447,48 @@ ${partial}` : "[INTERRUPTED] Tool execution was stopped.";
   }
 };
 
+// src/recording.ts
+function liftRecording(result) {
+  if (!result.includes('"recording"')) {
+    return { result };
+  }
+  let parsed;
+  try {
+    parsed = JSON.parse(result);
+  } catch {
+    return { result };
+  }
+  if (!parsed || typeof parsed !== "object" || Array.isArray(parsed)) {
+    return { result };
+  }
+  const obj = parsed;
+  const r = obj.recording;
+  if (!r || typeof r !== "object") {
+    return { result };
+  }
+  const ref = r;
+  if (typeof ref.path !== "string" || typeof ref.sessionId !== "string" || typeof ref.runId !== "string" || typeof ref.seq !== "number") {
+    return { result };
+  }
+  const recording = {
+    path: ref.path,
+    sessionId: ref.sessionId,
+    runId: ref.runId,
+    seq: ref.seq,
+    containsSnapshot: ref.containsSnapshot === true,
+    startTs: typeof ref.startTs === "number" ? ref.startTs : 0,
+    endTs: typeof ref.endTs === "number" ? ref.endTs : 0,
+    width: typeof ref.width === "number" ? ref.width : 0,
+    height: typeof ref.height === "number" ? ref.height : 0
+  };
+  const rest = { ...obj };
+  delete rest.recording;
+  return {
+    result: JSON.stringify({ ...rest, recorded: true }),
+    recording
+  };
+}
+
 // src/historyLimits.ts
 var MAX_TOOL_RESULT_BYTES = 256 * 1024;
 function capToolResult(result, maxBytes = MAX_TOOL_RESULT_BYTES) {
@@ -3478,6 +3531,13 @@ function capMessageForHistory(msg, maxBytes = MAX_TOOL_RESULT_BYTES) {
         continue;
       }
       if (typeof block.result === "string") {
+        if (block.name === "browserCommand" && !block.recording) {
+          const lifted = liftRecording(block.result);
+          if (lifted.recording) {
+            block.result = lifted.result;
+            block.recording = lifted.recording;
+          }
+        }
         block.result = capToolResult(block.result, maxBytes);
       }
       if (typeof block.backgroundResult === "string") {
@@ -4074,25 +4134,36 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
             }
             let settle;
             const resultPromise = new Promise((res) => {
-              settle = (result, isError) => res({ id: tc.id, result, isError });
+              settle = (result, isError, recording) => res({
+                id: tc.id,
+                result,
+                isError,
+                ...recording ? { recording } : {}
+              });
             });
             let toolAbort = new AbortController();
             const cascadeAbort = () => toolAbort.abort();
             signal?.addEventListener("abort", cascadeAbort, { once: true });
             let settled = false;
-            const safeSettle = (result, isError) => {
+            const safeSettle = (result, isError, recording) => {
               if (settled) {
                 return;
               }
               settled = true;
               signal?.removeEventListener("abort", cascadeAbort);
-              settle(result, isError);
+              settle(result, isError, recording);
             };
             const run2 = async (input) => {
               try {
                 let result;
+                let recording;
                 if (externalTools.has(tc.name) && resolveExternalTool) {
                   result = await resolveExternalTool(tc.id, tc.name, input);
+                  if (tc.name === "browserCommand") {
+                    const lifted = liftRecording(result);
+                    result = lifted.result;
+                    recording = lifted.recording;
+                  }
                 } else {
                   const onLog = (line) => emit({
                     type: "tool_input_delta",
@@ -4108,7 +4179,11 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
                     subAgentMessages
                   );
                 }
-                safeSettle(capToolResult(result), result.startsWith("Error"));
+                safeSettle(
+                  capToolResult(result),
+                  result.startsWith("Error"),
+                  recording
+                );
               } catch (err) {
                 safeSettle(`Error: ${err.message}`, true);
               }
@@ -4150,7 +4225,8 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
               id: tc.id,
               name: tc.name,
               result: r.result,
-              isError: r.isError
+              isError: r.isError,
+              ...r.recording ? { recording: r.recording } : {}
             });
             return r;
           })
@@ -4164,6 +4240,9 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
             block.result = r.result;
             block.isError = r.isError;
             block.completedAt = Date.now();
+            if (r.recording) {
+              block.recording = r.recording;
+            }
             const innerMsgs = subAgentMessages.get(r.id);
             if (innerMsgs) {
               block.subAgentMessages = capSubAgentTranscript(innerMsgs);
@@ -4294,7 +4373,7 @@ var BROWSER_TOOLS = [
   },
   {
     name: "browserCommand",
-    description: "Interact with the app's live preview by sending browser commands. Commands execute sequentially with an animated cursor. Always start with a snapshot to see the current state and get ref identifiers. The result includes a snapshot field with the final page state after all steps complete. On error, the failing step has an error field and execution stops. Batches that contain an interactive step (click, type, select) also return a `recording` object \u2014 one chunk of a continuous per-session rrweb recording that the viewer stitches into a single replay (not a standalone per-call clip). Timeout: 120s.",
+    description: "Interact with the app's live preview by sending browser commands. Commands execute sequentially with an animated cursor. Always start with a snapshot to see the current state and get ref identifiers. The result includes a snapshot field with the final page state after all steps complete. On error, the failing step has an error field and execution stops. Batches that contain an interactive step (click, type, select) are recorded for the user to replay and return `recorded: true`. Timeout: 120s.",
     inputSchema: {
       type: "object",
       properties: {
@@ -4412,7 +4491,7 @@ function walkMdFiles(dir, skip) {
     for (const entry of fs17.readdirSync(dir, { withFileTypes: true })) {
       const full = path9.join(dir, entry.name);
       if (entry.isDirectory()) {
-        if (!skip?.has(entry.name)) {
+        if (!skip?.has(entry.name) && !entry.name.startsWith(".")) {
           files.push(...walkMdFiles(full, skip));
         }
       } else if (entry.name.endsWith(".md")) {
@@ -4710,8 +4789,14 @@ async function runBrowserAutomation(task, context, opts) {
     });
     context.subAgentMessages?.set(context.toolCallId, result.messages);
     const preferred = opts?.capture === "viewport" ? lastCapture.viewport ?? lastCapture.fullPage : lastCapture.fullPage ?? lastCapture.viewport;
+    const recorded = result.messages.some(
+      (m) => m.role === "assistant" && Array.isArray(m.content) && m.content.some(
+        (b) => b.type === "tool" && b.name === "browserCommand" && !!b.recording
+      )
+    );
     return {
       text: result.text,
+      recorded,
       ...preferred?.url ? { screenshot: preferred } : {}
     };
   } finally {
@@ -4738,12 +4823,18 @@ var browserAutomationTool = {
       return "Error: browser automation requires execution context (only available in headless mode)";
     }
     const result = await runBrowserAutomation(input.task, context);
+    let text = result.text;
     if (result.screenshot) {
-      return `${result.text}
+      text += `
 
 ![Final state](${result.screenshot.url})`;
     }
-    return result.text;
+    if (result.recorded) {
+      text += `
+
+Replay of this run: ![Browser test replay](replay:${context.toolCallId})`;
+    }
+    return text;
   }
 };
 
@@ -7077,6 +7168,138 @@ After reconciling the spec, refresh the Build Overview: author the complete upda
   }
 };
 
+// src/subagents/reviewExistingProject/tools.ts
+var REVIEW_TOOLS = [
+  ...COMMON_READ_TOOLS,
+  bashTool.definition
+];
+
+// src/subagents/reviewExistingProject/validatePaths.ts
+import { stat as stat4 } from "fs/promises";
+import { join as join5 } from "path";
+var UPLOAD_REF_RE = /`(src\/\.user-uploads\/[^`\n]+?)\/?`/g;
+async function validateUploadPaths(text) {
+  const refs = /* @__PURE__ */ new Set();
+  for (const m of text.matchAll(UPLOAD_REF_RE)) {
+    refs.add(m[1]);
+  }
+  if (refs.size === 0) {
+    return null;
+  }
+  const missing = [];
+  for (const ref of refs) {
+    const exists = await stat4(join5(PROJECT_ROOT, ref)).then(
+      () => true,
+      () => false
+    );
+    if (!exists) {
+      missing.push(ref);
+    }
+  }
+  if (missing.length === 0) {
+    return null;
+  }
+  return [
+    `Your review cites paths under src/.user-uploads/ that do not exist on disk:`,
+    ...missing.map((p) => `- ${p}`),
+    ``,
+    `Check the real location with listDir \u2014 the trimmed tree may have landed under a different name, or the original archive may already have been deleted \u2014 and correct or remove each path. Then send your complete review again from the top \u2014 it fully replaces your previous response, so include everything, not just the fixes.`
+  ].join("\n");
+}
+
+// src/subagents/reviewExistingProject/index.ts
+var BASE_PROMPT7 = readAsset("subagents/reviewExistingProject", "prompt.md");
+function buildTask(input) {
+  const parts = [];
+  const paths = (input.paths ?? []).filter(
+    (p) => typeof p === "string" && p.trim()
+  );
+  if (paths.length > 0) {
+    parts.push(
+      `Uploads (local paths):
+${paths.map((p) => `- ${p.trim()}`).join("\n")}`
+    );
+  }
+  if (input.repoUrl?.trim()) {
+    parts.push(`Repository URL: ${input.repoUrl.trim()}`);
+  }
+  if (input.context?.trim()) {
+    parts.push(`Context from the conversation: ${input.context.trim()}`);
+  }
+  return parts.join("\n\n");
+}
+async function runReviewExistingProject(input, context) {
+  const task = buildTask(input);
+  if (!task) {
+    return "Error: reviewExistingProject needs at least one upload path or a repository URL.";
+  }
+  const specIndex = loadSpecIndex();
+  const parts = [BASE_PROMPT7, loadPlatformBrief()];
+  parts.push("<!-- cache_breakpoint -->");
+  if (specIndex) {
+    parts.push(specIndex);
+  }
+  const system = parts.join("\n\n");
+  const result = await runSubAgent({
+    system,
+    task,
+    tools: REVIEW_TOOLS,
+    externalTools: /* @__PURE__ */ new Set(),
+    executeTool: (name, toolInput, toolCallId, onLog, sams) => {
+      const childCtx = toolCallId ? {
+        ...deriveContext(context, toolCallId, onLog),
+        subAgentMessages: sams
+      } : context;
+      return executeTool(name, toolInput, childCtx);
+    },
+    apiConfig: context.apiConfig,
+    model: resolveModel("reviewExistingProject", context.models, context.model),
+    subAgentId: "reviewExistingProject",
+    signal: context.signal,
+    parentToolId: context.toolCallId,
+    requestId: context.requestId,
+    onEvent: context.onEvent,
+    resolveExternalTool: context.resolveExternalTool,
+    toolRegistry: context.toolRegistry,
+    // Every quoted src/.user-uploads/ path in the review must exist on disk
+    // (see validatePaths.ts) — the paths are the part the caller acts on.
+    validateResult: (text) => validateUploadPaths(text)
+  });
+  context.subAgentMessages?.set(context.toolCallId, result.messages);
+  return result.text;
+}
+var reviewExistingProjectTool = {
+  definition: {
+    name: "reviewExistingProject",
+    description: "Your reviewer for anything a user brings from a prior/extant project or attempt at a project: a zip of a codebase from another coding agent, an export from a vibe coding platform, a folder of specs and docs, a public git URL. It unpacks the upload, strips the noise, lands the trimmed tree in src/.user-uploads/, and returns a short review: what the project was trying to be, the materials worth carrying forward with their paths, what to ignore, state signals, and the questions only the user can answer. Treat the review as evidence of what the user wants, never as a spec. Do not open the archive yourself.",
+    inputSchema: {
+      type: "object",
+      properties: {
+        paths: {
+          type: "array",
+          items: { type: "string" },
+          description: "Local paths under src/.user-uploads/ to the uploaded archive(s) and/or loose files, exactly as given in the attachment header."
+        },
+        context: {
+          type: "string",
+          description: "One or two lines about the user and what they said they want. The reviewer cannot see the conversation."
+        },
+        repoUrl: {
+          type: "string",
+          description: "A public git repository URL to shallow-clone and review, instead of or alongside the uploads."
+        }
+      },
+      required: []
+    }
+  },
+  async execute(input, context) {
+    if (!context) {
+      return "Error: reviewExistingProject requires execution context";
+    }
+    return runReviewExistingProject(input, context);
+  }
+};
+
 // src/tools/common/scrapeWebUrl.ts
 var scrapeWebUrlTool = {
   definition: {
@@ -7171,7 +7394,8 @@ var ALL_TOOLS = [
   // new tool goes at the end to leave every existing session's prefix intact.
   loadSkillTool,
   testJewelTool,
-  researchTool
+  researchTool,
+  reviewExistingProjectTool
 ];
 var SUBAGENT_TOOL_NAMES = /* @__PURE__ */ new Set([
   "visualDesignExpert",
@@ -7181,7 +7405,8 @@ var SUBAGENT_TOOL_NAMES = /* @__PURE__ */ new Set([
   "specSync",
   "runAutomatedBrowserTest",
   "askMindStudioSdk",
-  "research"
+  "research",
+  "reviewExistingProject"
 ]);
 function getToolDefinitions() {
   return ALL_TOOLS.map((t) => t.definition);
@@ -8171,7 +8396,9 @@ function walkMdFiles2(dir) {
     for (const entry of entries) {
       const full = path12.join(dir, entry.name);
       if (entry.isDirectory()) {
-        results.push(...walkMdFiles2(full));
+        if (!entry.name.startsWith(".")) {
+          results.push(...walkMdFiles2(full));
+        }
       } else if (entry.name.endsWith(".md")) {
         results.push(full);
       }
@@ -9571,11 +9798,15 @@ async function runTurn(params) {
 }
 
 // src/headless/attachments.ts
-import { mkdirSync, existsSync } from "fs";
-import { writeFile as writeFile3 } from "fs/promises";
-import { basename as basename2, join as join5, extname as extname3 } from "path";
+import { mkdirSync, existsSync, createWriteStream } from "fs";
+import { writeFile as writeFile3, stat as stat5 } from "fs/promises";
+import { Readable } from "stream";
+import { pipeline } from "stream/promises";
+import { basename as basename2, join as join6, extname as extname3 } from "path";
 var log16 = createLogger("headless:attachments");
 var UPLOADS_DIR = "src/.user-uploads";
+var IMAGE_DOWNLOAD_TIMEOUT_MS = 3e4;
+var DOCUMENT_DOWNLOAD_TIMEOUT_MS = 3e5;
 function filenameFromUrl(url) {
   try {
     const pathname = new URL(url).pathname;
@@ -9586,7 +9817,7 @@ function filenameFromUrl(url) {
   }
 }
 function resolveUniqueFilename(name, claimed) {
-  const isFree = (candidate) => !claimed.has(candidate) && !existsSync(join5(UPLOADS_DIR, candidate));
+  const isFree = (candidate) => !claimed.has(candidate) && !existsSync(join6(UPLOADS_DIR, candidate));
   if (isFree(name)) {
     return name;
   }
@@ -9621,19 +9852,23 @@ async function persistAttachmentList(attachments) {
   const results = await Promise.allSettled(
     nonVoice.map(async (att, i) => {
       const name = names[i];
-      const localPath = join5(UPLOADS_DIR, name);
+      const localPath = join6(UPLOADS_DIR, name);
+      const timeoutMs = isImageAttachment(att) ? IMAGE_DOWNLOAD_TIMEOUT_MS : DOCUMENT_DOWNLOAD_TIMEOUT_MS;
       const res = await fetch(att.url, {
-        signal: AbortSignal.timeout(3e4)
+        signal: AbortSignal.timeout(timeoutMs)
       });
-      if (!res.ok) {
+      if (!res.ok || !res.body) {
         throw new Error(`HTTP ${res.status} downloading ${att.url}`);
       }
-      const buffer = Buffer.from(await res.arrayBuffer());
-      await writeFile3(localPath, buffer);
+      await pipeline(
+        Readable.fromWeb(res.body),
+        createWriteStream(localPath)
+      );
+      const { size } = await stat5(localPath);
       log16.info("Attachment saved", {
         filename: name,
         path: localPath,
-        bytes: buffer.length
+        bytes: size
       });
       let extractedTextPath;
       if (att.extractedTextUrl) {
@@ -10504,6 +10739,7 @@ var HeadlessSession = class {
             name: e.name,
             result: e.result,
             isError: e.isError,
+            ...e.recording && { recording: e.recording },
             ...e.parentToolId && { parentToolId: e.parentToolId }
           },
           rid

package/dist/index.js

@@ -824,6 +824,9 @@ async function listRecursive(dir) {
   for (const entry of entries) {
     const fullPath = path2.join(dir, entry.name);
     if (entry.isDirectory()) {
+      if (entry.name.startsWith(".")) {
+        continue;
+      }
       results.push(`${fullPath}/`);
       results.push(...await listRecursive(fullPath));
     } else {
@@ -1947,7 +1950,7 @@ var init_compaction = __esm({
     init_api();
     init_assets();
     init_sentinel();
-    init_tools9();
+    init_tools10();
     init_logger();
     init_usageLedger();
     log2 = createLogger("compaction");
@@ -2058,6 +2061,13 @@ var init_surfaces = __esm({
         modelType: "text",
         userPickable: true
       },
+      reviewExistingProject: {
+        default: "claude-5-sonnet",
+        label: "Existing Project Review",
+        description: "Reviews a project you bring from another tool and reports what is worth carrying forward.",
+        modelType: "text",
+        userPickable: true
+      },
       copyEditor: {
         default: "claude-5-sonnet",
         label: "Copy Agent",
@@ -2129,6 +2139,7 @@ var init_surfaces = __esm({
       "gpt-5.6-sol": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
       "gpt-5.6-terra": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
       "gpt-5.6-luna": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
+      "gpt-6-astra": { forceCompactAt: 6e5, suggestCompactAt: 25e4 },
       // Google ~1M-context; only 3.1-pro is tiered (higher rates above 200K).
       "gemini-3-pro": { forceCompactAt: 85e4 },
       "gemini-3.1-pro": { forceCompactAt: 85e4, suggestCompactAt: 18e4 },
@@ -2331,6 +2342,53 @@ var init_cleanMessages = __esm({
   }
 });
 
+// src/recording.ts
+function liftRecording(result) {
+  if (!result.includes('"recording"')) {
+    return { result };
+  }
+  let parsed;
+  try {
+    parsed = JSON.parse(result);
+  } catch {
+    return { result };
+  }
+  if (!parsed || typeof parsed !== "object" || Array.isArray(parsed)) {
+    return { result };
+  }
+  const obj = parsed;
+  const r = obj.recording;
+  if (!r || typeof r !== "object") {
+    return { result };
+  }
+  const ref = r;
+  if (typeof ref.path !== "string" || typeof ref.sessionId !== "string" || typeof ref.runId !== "string" || typeof ref.seq !== "number") {
+    return { result };
+  }
+  const recording = {
+    path: ref.path,
+    sessionId: ref.sessionId,
+    runId: ref.runId,
+    seq: ref.seq,
+    containsSnapshot: ref.containsSnapshot === true,
+    startTs: typeof ref.startTs === "number" ? ref.startTs : 0,
+    endTs: typeof ref.endTs === "number" ? ref.endTs : 0,
+    width: typeof ref.width === "number" ? ref.width : 0,
+    height: typeof ref.height === "number" ? ref.height : 0
+  };
+  const rest = { ...obj };
+  delete rest.recording;
+  return {
+    result: JSON.stringify({ ...rest, recorded: true }),
+    recording
+  };
+}
+var init_recording = __esm({
+  "src/recording.ts"() {
+    "use strict";
+  }
+});
+
 // src/historyLimits.ts
 function capToolResult(result, maxBytes = MAX_TOOL_RESULT_BYTES) {
   const total = Buffer.byteLength(result, "utf-8");
@@ -2370,6 +2428,13 @@ function capMessageForHistory(msg, maxBytes = MAX_TOOL_RESULT_BYTES) {
         continue;
       }
       if (typeof block.result === "string") {
+        if (block.name === "browserCommand" && !block.recording) {
+          const lifted = liftRecording(block.result);
+          if (lifted.recording) {
+            block.result = lifted.result;
+            block.recording = lifted.recording;
+          }
+        }
         block.result = capToolResult(block.result, maxBytes);
       }
       if (typeof block.backgroundResult === "string") {
@@ -2390,6 +2455,7 @@ var MAX_TOOL_RESULT_BYTES, MAX_SUBAGENT_RESULT_BYTES, MAX_SUBAGENT_TRANSCRIPT_BY
 var init_historyLimits = __esm({
   "src/historyLimits.ts"() {
     "use strict";
+    init_recording();
     MAX_TOOL_RESULT_BYTES = 256 * 1024;
     MAX_SUBAGENT_RESULT_BYTES = 32 * 1024;
     MAX_SUBAGENT_TRANSCRIPT_BYTES = 512 * 1024;
@@ -3787,8 +3853,8 @@ function formatSize(bytes) {
 }
 async function formatFile(dirPath, name, indent) {
   try {
-    const stat4 = await fs15.stat(path8.join(dirPath, name));
-    return `${indent}${name}${" ".repeat(Math.max(1, 30 - indent.length - name.length))}${formatSize(stat4.size)}`;
+    const stat6 = await fs15.stat(path8.join(dirPath, name));
+    return `${indent}${name}${" ".repeat(Math.max(1, 30 - indent.length - name.length))}${formatSize(stat6.size)}`;
   } catch {
     return `${indent}${name}`;
   }
@@ -5092,25 +5158,36 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
             }
             let settle;
             const resultPromise = new Promise((res) => {
-              settle = (result, isError) => res({ id: tc.id, result, isError });
+              settle = (result, isError, recording) => res({
+                id: tc.id,
+                result,
+                isError,
+                ...recording ? { recording } : {}
+              });
             });
             let toolAbort = new AbortController();
             const cascadeAbort = () => toolAbort.abort();
             signal?.addEventListener("abort", cascadeAbort, { once: true });
             let settled = false;
-            const safeSettle = (result, isError) => {
+            const safeSettle = (result, isError, recording) => {
               if (settled) {
                 return;
               }
               settled = true;
               signal?.removeEventListener("abort", cascadeAbort);
-              settle(result, isError);
+              settle(result, isError, recording);
             };
             const run2 = async (input) => {
               try {
                 let result;
+                let recording;
                 if (externalTools.has(tc.name) && resolveExternalTool) {
                   result = await resolveExternalTool(tc.id, tc.name, input);
+                  if (tc.name === "browserCommand") {
+                    const lifted = liftRecording(result);
+                    result = lifted.result;
+                    recording = lifted.recording;
+                  }
                 } else {
                   const onLog = (line) => emit({
                     type: "tool_input_delta",
@@ -5126,7 +5203,11 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
                     subAgentMessages
                   );
                 }
-                safeSettle(capToolResult(result), result.startsWith("Error"));
+                safeSettle(
+                  capToolResult(result),
+                  result.startsWith("Error"),
+                  recording
+                );
               } catch (err) {
                 safeSettle(`Error: ${err.message}`, true);
               }
@@ -5168,7 +5249,8 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
               id: tc.id,
               name: tc.name,
               result: r.result,
-              isError: r.isError
+              isError: r.isError,
+              ...r.recording ? { recording: r.recording } : {}
             });
             return r;
           })
@@ -5182,6 +5264,9 @@ ${partial}` : "[INTERRUPTED] Agent was interrupted before producing output.",
             block.result = r.result;
             block.isError = r.isError;
             block.completedAt = Date.now();
+            if (r.recording) {
+              block.recording = r.recording;
+            }
             const innerMsgs = subAgentMessages.get(r.id);
             if (innerMsgs) {
               block.subAgentMessages = capSubAgentTranscript(innerMsgs);
@@ -5273,6 +5358,7 @@ var init_runner = __esm({
     init_usageLedger();
     init_toolRegistry();
     init_historyLimits();
+    init_recording();
     init_statusWatcher();
     init_cleanMessages();
     log8 = createLogger("sub-agent");
@@ -5342,7 +5428,7 @@ var init_tools2 = __esm({
       },
       {
         name: "browserCommand",
-        description: "Interact with the app's live preview by sending browser commands. Commands execute sequentially with an animated cursor. Always start with a snapshot to see the current state and get ref identifiers. The result includes a snapshot field with the final page state after all steps complete. On error, the failing step has an error field and execution stops. Batches that contain an interactive step (click, type, select) also return a `recording` object \u2014 one chunk of a continuous per-session rrweb recording that the viewer stitches into a single replay (not a standalone per-call clip). Timeout: 120s.",
+        description: "Interact with the app's live preview by sending browser commands. Commands execute sequentially with an animated cursor. Always start with a snapshot to see the current state and get ref identifiers. The result includes a snapshot field with the final page state after all steps complete. On error, the failing step has an error field and execution stops. Batches that contain an interactive step (click, type, select) are recorded for the user to replay and return `recorded: true`. Timeout: 120s.",
         inputSchema: {
           type: "object",
           properties: {
@@ -5462,7 +5548,7 @@ function walkMdFiles(dir, skip) {
     for (const entry of fs16.readdirSync(dir, { withFileTypes: true })) {
       const full = path9.join(dir, entry.name);
       if (entry.isDirectory()) {
-        if (!skip?.has(entry.name)) {
+        if (!skip?.has(entry.name) && !entry.name.startsWith(".")) {
           files.push(...walkMdFiles(full, skip));
         }
       } else if (entry.name.endsWith(".md")) {
@@ -5771,8 +5857,14 @@ async function runBrowserAutomation(task, context, opts) {
     });
     context.subAgentMessages?.set(context.toolCallId, result.messages);
     const preferred = opts?.capture === "viewport" ? lastCapture.viewport ?? lastCapture.fullPage : lastCapture.fullPage ?? lastCapture.viewport;
+    const recorded = result.messages.some(
+      (m) => m.role === "assistant" && Array.isArray(m.content) && m.content.some(
+        (b) => b.type === "tool" && b.name === "browserCommand" && !!b.recording
+      )
+    );
     return {
       text: result.text,
+      recorded,
       ...preferred?.url ? { screenshot: preferred } : {}
     };
   } finally {
@@ -5783,7 +5875,7 @@ var log9, CAPTURE_COMMANDS, browserAutomationTool;
 var init_browserAutomation = __esm({
   "src/subagents/browserAutomation/index.ts"() {
     "use strict";
-    init_tools9();
+    init_tools10();
     init_runner();
     init_tools2();
     init_tools();
@@ -5817,12 +5909,18 @@ var init_browserAutomation = __esm({
           return "Error: browser automation requires execution context (only available in headless mode)";
         }
         const result = await runBrowserAutomation(input.task, context);
+        let text = result.text;
         if (result.screenshot) {
-          return `${result.text}
+          text += `
 
 ![Final state](${result.screenshot.url})`;
         }
-        return result.text;
+        if (result.recorded) {
+          text += `
+
+Replay of this run: ![Browser test replay](replay:${context.toolCallId})`;
+        }
+        return text;
       }
     };
   }
@@ -6081,7 +6179,7 @@ var init_research = __esm({
     init_assets();
     init_runner();
     init_context();
-    init_tools9();
+    init_tools10();
     init_tools3();
     init_surfaces();
     BASE_PROMPT2 = readAsset("subagents/research", "prompt.md");
@@ -6949,7 +7047,7 @@ var init_copyEditor = __esm({
     init_assets();
     init_runner();
     init_context();
-    init_tools9();
+    init_tools10();
     init_tools4();
     init_surfaces();
     BASE_PROMPT3 = readAsset("subagents/copyEditor", "prompt.md");
@@ -7279,7 +7377,7 @@ var research, tools, DESIGN_EXPERT_TOOLS;
 var init_tools5 = __esm({
   "src/subagents/designExpert/tools/index.ts"() {
     "use strict";
-    init_tools9();
+    init_tools10();
     init_tools();
     init_research();
     init_scrapeWebUrl();
@@ -7808,7 +7906,7 @@ var DESCRIPTION, RENDER_WRITE_TOOL_NAMES, DESIGN_EXPERT_RENDER_TOOLS, designExpe
 var init_designExpert = __esm({
   "src/subagents/designExpert/index.ts"() {
     "use strict";
-    init_tools9();
+    init_tools10();
     init_runner();
     init_tools5();
     init_tools();
@@ -8061,7 +8159,7 @@ var productVisionTool;
 var init_productVision = __esm({
   "src/subagents/productVision/index.ts"() {
     "use strict";
-    init_tools9();
+    init_tools10();
     init_runner();
     init_tools6();
     init_tools();
@@ -8216,7 +8314,7 @@ var init_codeSanityCheck = __esm({
     init_assets();
     init_runner();
     init_context();
-    init_tools9();
+    init_tools10();
     init_tools7();
     init_research();
     init_tools3();
@@ -8453,7 +8551,7 @@ var init_specSync = __esm({
     init_assets();
     init_runner();
     init_context();
-    init_tools9();
+    init_tools10();
     init_writeBuildOverview();
     init_tools8();
     init_lock();
@@ -8559,6 +8657,166 @@ After reconciling the spec, refresh the Build Overview: author the complete upda
   }
 });
 
+// src/subagents/reviewExistingProject/tools.ts
+var REVIEW_TOOLS;
+var init_tools9 = __esm({
+  "src/subagents/reviewExistingProject/tools.ts"() {
+    "use strict";
+    init_tools();
+    init_bash();
+    REVIEW_TOOLS = [
+      ...COMMON_READ_TOOLS,
+      bashTool.definition
+    ];
+  }
+});
+
+// src/subagents/reviewExistingProject/validatePaths.ts
+import { stat as stat4 } from "fs/promises";
+import { join as join5 } from "path";
+async function validateUploadPaths(text) {
+  const refs = /* @__PURE__ */ new Set();
+  for (const m of text.matchAll(UPLOAD_REF_RE)) {
+    refs.add(m[1]);
+  }
+  if (refs.size === 0) {
+    return null;
+  }
+  const missing = [];
+  for (const ref of refs) {
+    const exists = await stat4(join5(PROJECT_ROOT, ref)).then(
+      () => true,
+      () => false
+    );
+    if (!exists) {
+      missing.push(ref);
+    }
+  }
+  if (missing.length === 0) {
+    return null;
+  }
+  return [
+    `Your review cites paths under src/.user-uploads/ that do not exist on disk:`,
+    ...missing.map((p) => `- ${p}`),
+    ``,
+    `Check the real location with listDir \u2014 the trimmed tree may have landed under a different name, or the original archive may already have been deleted \u2014 and correct or remove each path. Then send your complete review again from the top \u2014 it fully replaces your previous response, so include everything, not just the fixes.`
+  ].join("\n");
+}
+var UPLOAD_REF_RE;
+var init_validatePaths = __esm({
+  "src/subagents/reviewExistingProject/validatePaths.ts"() {
+    "use strict";
+    init_projectRoot();
+    UPLOAD_REF_RE = /`(src\/\.user-uploads\/[^`\n]+?)\/?`/g;
+  }
+});
+
+// src/subagents/reviewExistingProject/index.ts
+function buildTask(input) {
+  const parts = [];
+  const paths = (input.paths ?? []).filter(
+    (p) => typeof p === "string" && p.trim()
+  );
+  if (paths.length > 0) {
+    parts.push(
+      `Uploads (local paths):
+${paths.map((p) => `- ${p.trim()}`).join("\n")}`
+    );
+  }
+  if (input.repoUrl?.trim()) {
+    parts.push(`Repository URL: ${input.repoUrl.trim()}`);
+  }
+  if (input.context?.trim()) {
+    parts.push(`Context from the conversation: ${input.context.trim()}`);
+  }
+  return parts.join("\n\n");
+}
+async function runReviewExistingProject(input, context) {
+  const task = buildTask(input);
+  if (!task) {
+    return "Error: reviewExistingProject needs at least one upload path or a repository URL.";
+  }
+  const specIndex = loadSpecIndex();
+  const parts = [BASE_PROMPT7, loadPlatformBrief()];
+  parts.push("<!-- cache_breakpoint -->");
+  if (specIndex) {
+    parts.push(specIndex);
+  }
+  const system = parts.join("\n\n");
+  const result = await runSubAgent({
+    system,
+    task,
+    tools: REVIEW_TOOLS,
+    externalTools: /* @__PURE__ */ new Set(),
+    executeTool: (name, toolInput, toolCallId, onLog, sams) => {
+      const childCtx = toolCallId ? {
+        ...deriveContext(context, toolCallId, onLog),
+        subAgentMessages: sams
+      } : context;
+      return executeTool(name, toolInput, childCtx);
+    },
+    apiConfig: context.apiConfig,
+    model: resolveModel("reviewExistingProject", context.models, context.model),
+    subAgentId: "reviewExistingProject",
+    signal: context.signal,
+    parentToolId: context.toolCallId,
+    requestId: context.requestId,
+    onEvent: context.onEvent,
+    resolveExternalTool: context.resolveExternalTool,
+    toolRegistry: context.toolRegistry,
+    // Every quoted src/.user-uploads/ path in the review must exist on disk
+    // (see validatePaths.ts) — the paths are the part the caller acts on.
+    validateResult: (text) => validateUploadPaths(text)
+  });
+  context.subAgentMessages?.set(context.toolCallId, result.messages);
+  return result.text;
+}
+var BASE_PROMPT7, reviewExistingProjectTool;
+var init_reviewExistingProject = __esm({
+  "src/subagents/reviewExistingProject/index.ts"() {
+    "use strict";
+    init_assets();
+    init_runner();
+    init_context();
+    init_tools10();
+    init_tools9();
+    init_validatePaths();
+    init_surfaces();
+    BASE_PROMPT7 = readAsset("subagents/reviewExistingProject", "prompt.md");
+    reviewExistingProjectTool = {
+      definition: {
+        name: "reviewExistingProject",
+        description: "Your reviewer for anything a user brings from a prior/extant project or attempt at a project: a zip of a codebase from another coding agent, an export from a vibe coding platform, a folder of specs and docs, a public git URL. It unpacks the upload, strips the noise, lands the trimmed tree in src/.user-uploads/, and returns a short review: what the project was trying to be, the materials worth carrying forward with their paths, what to ignore, state signals, and the questions only the user can answer. Treat the review as evidence of what the user wants, never as a spec. Do not open the archive yourself.",
+        inputSchema: {
+          type: "object",
+          properties: {
+            paths: {
+              type: "array",
+              items: { type: "string" },
+              description: "Local paths under src/.user-uploads/ to the uploaded archive(s) and/or loose files, exactly as given in the attachment header."
+            },
+            context: {
+              type: "string",
+              description: "One or two lines about the user and what they said they want. The reviewer cannot see the conversation."
+            },
+            repoUrl: {
+              type: "string",
+              description: "A public git repository URL to shallow-clone and review, instead of or alongside the uploads."
+            }
+          },
+          required: []
+        }
+      },
+      async execute(input, context) {
+        if (!context) {
+          return "Error: reviewExistingProject requires execution context";
+        }
+        return runReviewExistingProject(input, context);
+      }
+    };
+  }
+});
+
 // src/tools/common/scrapeWebUrl.ts
 var scrapeWebUrlTool;
 var init_scrapeWebUrl2 = __esm({
@@ -8629,7 +8887,7 @@ function executeTool(name, input, context) {
   return tool.execute(input, context);
 }
 var ALL_TOOLS, SUBAGENT_TOOL_NAMES;
-var init_tools9 = __esm({
+var init_tools10 = __esm({
   "src/tools/index.ts"() {
     "use strict";
     init_readSpec();
@@ -8668,6 +8926,7 @@ var init_tools9 = __esm({
     init_copyEditor();
     init_specSync();
     init_research();
+    init_reviewExistingProject();
     init_scrapeWebUrl2();
     init_writeBuildOverview();
     ALL_TOOLS = [
@@ -8716,7 +8975,8 @@ var init_tools9 = __esm({
       // new tool goes at the end to leave every existing session's prefix intact.
       loadSkillTool,
       testJewelTool,
-      researchTool
+      researchTool,
+      reviewExistingProjectTool
     ];
     SUBAGENT_TOOL_NAMES = /* @__PURE__ */ new Set([
       "visualDesignExpert",
@@ -8726,7 +8986,8 @@ var init_tools9 = __esm({
       "specSync",
       "runAutomatedBrowserTest",
       "askMindStudioSdk",
-      "research"
+      "research",
+      "reviewExistingProject"
     ]);
   }
 });
@@ -9243,7 +9504,9 @@ function walkMdFiles2(dir) {
     for (const entry of entries) {
       const full = path11.join(dir, entry.name);
       if (entry.isDirectory()) {
-        results.push(...walkMdFiles2(full));
+        if (!entry.name.startsWith(".")) {
+          results.push(...walkMdFiles2(full));
+        }
       } else if (entry.name.endsWith(".md")) {
         results.push(full);
       }
@@ -10238,7 +10501,7 @@ var init_agent = __esm({
   "src/agent.ts"() {
     "use strict";
     init_api();
-    init_tools9();
+    init_tools10();
     init_session();
     init_logger();
     init_usageLedger();
@@ -10525,9 +10788,11 @@ var init_config = __esm({
 });
 
 // src/headless/attachments.ts
-import { mkdirSync, existsSync } from "fs";
-import { writeFile as writeFile3 } from "fs/promises";
-import { basename as basename2, join as join5, extname as extname3 } from "path";
+import { mkdirSync, existsSync, createWriteStream } from "fs";
+import { writeFile as writeFile3, stat as stat5 } from "fs/promises";
+import { Readable } from "stream";
+import { pipeline } from "stream/promises";
+import { basename as basename2, join as join6, extname as extname3 } from "path";
 function filenameFromUrl(url) {
   try {
     const pathname = new URL(url).pathname;
@@ -10538,7 +10803,7 @@ function filenameFromUrl(url) {
   }
 }
 function resolveUniqueFilename(name, claimed) {
-  const isFree = (candidate) => !claimed.has(candidate) && !existsSync(join5(UPLOADS_DIR, candidate));
+  const isFree = (candidate) => !claimed.has(candidate) && !existsSync(join6(UPLOADS_DIR, candidate));
   if (isFree(name)) {
     return name;
   }
@@ -10572,19 +10837,23 @@ async function persistAttachmentList(attachments) {
   const results = await Promise.allSettled(
     nonVoice.map(async (att, i) => {
       const name = names[i];
-      const localPath = join5(UPLOADS_DIR, name);
+      const localPath = join6(UPLOADS_DIR, name);
+      const timeoutMs = isImageAttachment(att) ? IMAGE_DOWNLOAD_TIMEOUT_MS : DOCUMENT_DOWNLOAD_TIMEOUT_MS;
       const res = await fetch(att.url, {
-        signal: AbortSignal.timeout(3e4)
+        signal: AbortSignal.timeout(timeoutMs)
       });
-      if (!res.ok) {
+      if (!res.ok || !res.body) {
         throw new Error(`HTTP ${res.status} downloading ${att.url}`);
       }
-      const buffer = Buffer.from(await res.arrayBuffer());
-      await writeFile3(localPath, buffer);
+      await pipeline(
+        Readable.fromWeb(res.body),
+        createWriteStream(localPath)
+      );
+      const { size } = await stat5(localPath);
       log16.info("Attachment saved", {
         filename: name,
         path: localPath,
-        bytes: buffer.length
+        bytes: size
       });
       let extractedTextPath;
       if (att.extractedTextUrl) {
@@ -10644,13 +10913,15 @@ function buildUploadHeader(documents, images) {
   return `[Uploaded files]
 ${lines.join("\n")}`;
 }
-var log16, UPLOADS_DIR, IMAGE_EXTENSIONS;
+var log16, UPLOADS_DIR, IMAGE_DOWNLOAD_TIMEOUT_MS, DOCUMENT_DOWNLOAD_TIMEOUT_MS, IMAGE_EXTENSIONS;
 var init_attachments = __esm({
   "src/headless/attachments.ts"() {
     "use strict";
     init_logger();
     log16 = createLogger("headless:attachments");
     UPLOADS_DIR = "src/.user-uploads";
+    IMAGE_DOWNLOAD_TIMEOUT_MS = 3e4;
+    DOCUMENT_DOWNLOAD_TIMEOUT_MS = 3e5;
     IMAGE_EXTENSIONS = /* @__PURE__ */ new Set([".png", ".jpg", ".jpeg", ".gif", ".webp"]);
   }
 });
@@ -10954,7 +11225,7 @@ var init_headless = __esm({
     init_attachments();
     init_planFile();
     init_stats();
-    init_tools9();
+    init_tools10();
     init_messageQueue();
     init_resolve();
     init_sentinel();
@@ -11512,6 +11783,7 @@ var init_headless = __esm({
                 name: e.name,
                 result: e.result,
                 isError: e.isError,
+                ...e.recording && { recording: e.recording },
                 ...e.parentToolId && { parentToolId: e.parentToolId }
               },
               rid

package/dist/prompt/static/intake.md

@@ -55,7 +55,7 @@ But know when to stop exploring. Once there's a clear concept with a specific au
 2. **Structured forms** — Use `promptUser` with `type: "form"` to collect details. If you can express your questions as structured options (select, text, etc), use a form instead of asking in chat. Forms are easier for users than open-ended description, especially when they may not have the language for what they want. Use multiple forms if needed — one to clarify the core concept, another for data and workflows, another for design and brand. Each form should build on what you've already learned. Always use `type: "form"` during intake.
 3. **Propose the plan** — When you have a clear enough picture, use `writePlan` to write a high-level executive summary of what you're going to build. This is not a technical spec or a file list. Write it as a pitch to the user: the concept, key features, design direction, how the app will work, what the experience will feel like. The user can approve, discuss, or reject. Do NOT start writing spec files or code until the plan is approved.
 
-During intake, the only tools you should use are `promptUser` (for forms) and `writePlan` (to propose the plan). Do not call sub-agents, write files, or do any other work. Intake is about understanding the user and proposing what to build. Everything else happens after the plan is approved.
+During intake, the only tools you should use are `promptUser` (for forms), `writePlan` (to propose the plan), and `reviewExistingProject` when the user has brought an existing project. Do not call other sub-agents, write files, or do any other work. Intake is about understanding the user and proposing what to build. Everything else happens after the plan is approved.
 
 ### What NOT to Do
 

package/dist/prompt/static/team.md

@@ -44,6 +44,12 @@ Send it anything where the answer lives outside this project — objective quest
 
 Brief it neutrally: state the question and any concrete context, do not lead it by hinting at the answer you expect. It is tasked with testing your assumptions, and it will tell you when the evidence says you're wrong; that report is the valuable one, so don't tilt it. You still have `scrapeWebUrl` for directly reading a URL the user gives you — that's fetching, not research.
 
+### Existing Project Review (`reviewExistingProject`)
+
+Your reviewer for anything a user brings from a prior/extant project or attempt at a project: a zip of a codebase from another coding agent, an export from a vibe coding platform, a folder of specs and docs, a public git URL. It unpacks the upload itself in `src/.user-uploads/`, strips the noise, and returns a short review: what the project was trying to be, the materials worth carrying forward with their paths, what to ignore, state signals, and the questions only the user can answer.
+
+Send it the raw upload paths and a line of context about the user. Treat the review as evidence about what the user wants, the same as a deck or a transcript, never as a spec. Its questions belong in your intake form. It is the one sub-agent you may call during intake, because that is when it is needed.
+
 ### Copy Agent (`copyEditor`)
 
 Your editor — a design expert for words. Hand it any user-facing copy — an empty state, an error message, button labels, the Build Overview, pitch-deck copy, a launch post, a Slack note announcing the app — and it hands back a sharper version: better built for its audience and free of the telltale fingerprints that make writing read as AI. You're good at deciding *what* to say; it's great at making it land. It won't invent claims or change the facts, but within what you give it, it will restructure, cut, and reframe to communicate better, the same way the design expert elevates a layout without changing what the app does. Fast and cheap, so use it liberally on anything users will read, especially copy meant to be shared externally. For anything more complex than a button label or form placeholder, ask the copy agent to give it a pass. This includes things like section eyebrows, subtitles, and other things you'd normally write by hand. Give it the text plus what it's for (the medium, the audience). Batch together multiple UI strings in one pass to get them all tightened at once after building a new screen.
@@ -58,7 +64,7 @@ It can also keep the Build Overview current: set `refreshBuildOverview: true` an
 
 ### QA (`runAutomatedBrowserTest`)
 
-For verifying complex stateful interactions: multi-step form submissions, auth flows, real-time updates, flows that require specific data/role setup. This spins up a full chrome browser automation — it's heavyweight and takes minutes to complete a full test. Do not use it for basic rendering or navigation checks. If you can verify something with a screenshot or by reading the code, do that instead. Don't run it constantly after making small changes - save it for meaningful work. Run a scenario first to seed test data and set user roles. The user is able to watch QA work on their screen via a live browser preview - the cursor will move, type, etc - so you can also use this to demo functionality to the user and help them understand how to use their app.
+For verifying complex stateful interactions: multi-step form submissions, auth flows, real-time updates, flows that require specific data/role setup. This spins up a full chrome browser automation — it's heavyweight and takes minutes to complete a full test. Do not use it for basic rendering or navigation checks. If you can verify something with a screenshot or by reading the code, do that instead. Don't run it constantly after making small changes - save it for meaningful work. Run a scenario first to seed test data and set user roles. The user is able to watch QA work on their screen via a live browser preview - the cursor will move, type, etc - so you can also use this to demo functionality to the user and help them understand how to use their app. A recorded run's result ends with a replay reference line: paste it verbatim into your reply when the run itself is what the user should see (they asked to watch it, or the replay shows a failure better than words), and skip it for routine checks.
 
 The QA agent can see the screen. Describe what to test, not how — it will figure out what to click, what to check, and what values to use. By default, it always starts its tests logged out/unauthenticated on "/" root, but if you want to test a deeper piece of the app it can bypass auth and automatically authenticate itself as any user/role - just tell it to authenticate as the test user and navigate to X to start the test. If a specific auth role is required to access the content, be sure to note that - it can automatically assume it for the purpose of testing. After every test session, the browser is reset to / and any authentication used or created by the tester is cleared and reset.
 

package/dist/subagents/browserAutomation/prompt.md

@@ -80,7 +80,7 @@ Each browserCommand returns:
 - `snapshot`: the final page state after all steps complete (always present, even without an explicit snapshot step)
 - `logs`: array of browser-side events that fired during the batch (console output, network failures, JS errors, user interactions). Check this for errors before reporting pass.
 - `duration`: total execution time in ms
-- `recording` (optional): metadata for an rrweb session recording, present whenever the batch contained an interactive step (click, type, select). Each call returns one chunk of a continuous per-session recording (the viewer stitches chunks by `sessionId`/`seq` into a single replay) — it's not a standalone clip. Note in your failure reports that a recording is available so the main agent can surface it.
+- `recorded` (optional): `true` when the batch contained an interactive step (click, type, select) and a replay of it was saved. The editor stitches every recorded batch of the session into one continuous replay the user can watch. Note in your failure reports that a recording is available so the main agent can surface it.
 
 On error, the failing step has an `error` field and execution stops. Remaining steps are skipped.
 

package/dist/subagents/reviewExistingProject/prompt.md

@@ -0,0 +1,43 @@
+You review an existing software project that a user has brought to be built on Remy. It might be a zip of a codebase from another coding agent (Claude Code, Codex, Cursor), an export from a builder (Lovable, Replit, Bolt, v0), a folder of specs and docs, a public git repository, or a mix. It might be a finsihed project, but, more than likely, the user usually could not finish it or could not run it. Your teammate is about to run the intake conversation with the user and needs to know what, if anything, in the project actually matters.
+
+Your job is to find what matters for building the user's app here. You are not cataloguing the project, auditing it, or planning a port. The prior attempt is evidence of what the user wants, the same as a deck or a meeting transcript. Most of it is the previous tool's work, not the user's, and it should not shape what gets built.
+
+Remember, this is a **communication shortcut** to more quickly align with a user on their goals and save time in defining an MVP, NOT a "project import" feature!
+
+## What to look for
+
+**Worth carrying forward** (report these, each with its path):
+
+- Intent: what the project was trying to be. README, docs, decks, transcripts, notes, and the prose in the previous agent's instruction files. CLAUDE.md and its relatives often hold the best product description in the whole tree.
+- Prompts and domain knowledge: system prompts, rule sets, reference tables, anything actually encoding the user's expertise - NOT things that were generated by a coding agent.
+- Real inputs and data: datasets, sample documents, fixtures that look like real customer material, expected outputs.
+- Brand: design tokens, fonts, logos, style guides, an existing look the user may want to keep.
+- Copy and content the user wrote (NOT ai-generated copy)
+- The domain's entities, read from the data model as evidence of what the app is about, not as a schema to reproduce. Do *NOT* overindex on extant schemas - let Remy re-discover schemas in a way that makes sense as it thinks through the build and aligns with the user, without being constrained to anything this early.
+- The user's own backlog or wishlist, when it is clearly theirs and not the tool's.
+
+**Not worth carrying forward**: tests, harnesses, CI, deployment and infra docs, lockfiles, dependencies, infrastructure choices, package choices, frameworks, tools, build output, generated boilerplate, and the previous agent's config and instruction files as instructions. Skim their prose for intent; inherit none of their decisions. Everything is green field from here on out.
+
+## Important Caveat
+
+The code in this project was more-than-likely written by a frontier coding agent. It very likely looks *immaculate* and world-class. There are probably whole hosts of tests, specs, documentation, claude.md, agents.md - the works. It scans like a robust, professional, software project. This does NOT, however, mean that it is aligned with what the user wants. The coding agent that made it did its job well - but clearly there is a reason the user is bringing their project to Remy :) Be extremeley careful not to mistake robustness, depth, and maturity of the *codebase* for robustness, depth, and maturity of the *product* or the *idea*. Just because there are a ton of tests doesn't mean that's what the user actually wants or needs! The 99% likely case is that the coding agent overbuilt and the user got in over their head. **Please be sure to keep this tension at the top of your mind while you review the project!**
+
+## How to work
+
+- Unzip each archive into a scratch directory under `/tmp/review/`. For a repository URL, `git clone --depth 1` into the same scratch directory.
+- In the scratch copy delete `node_modules`, `.git`, build output, caches, and lockfiles. If you find a real secret (a populated `.env`, keys, tokens), move it to a `_quarantine/` folder inside the tree and say so in the review.
+- Move the trimmed tree to `src/.user-uploads/<name>/`, where `<name>` is a short lowercase hyphenated name derived from the archive or repository (for example `pas-anomaly-app`). Then delete the original archive from `src/.user-uploads/`; the trimmed tree is now the durable copy and persists with the project. Loose documents that were uploaded on their own stay where they are.
+- Recognize the source by its shape and go straight to where the useful material lives. Read selectively and stop when the picture is clear; most reviews need fifteen to twenty-five tool calls. Fire independent reads together.
+- Never modify, install, build, or run anything outside `/tmp/review/` and `src/.user-uploads/`.
+
+## What you return
+
+Return a robust report in Markdown, including things like:
+- What was the user trying to make
+- What ideas/things/resources are worth carrying forward (include paths if there is a specific file, for example)
+- Summarize any other signals of intent like human-authored todos, a backlog, roadmap notes, etc
+- Where things are on the disk, including the trimmed tree's path, any quarantined files, anything you could not unpack
+- Questions only the user can answer to help disambugate (e.g., whether the data is real material or was generated by an agent, whether they want the same look/feel or something new, what they still want, etc).
+- Anything else that feels relevant to understanding the user's intent.
+
+Do not propose architecture, write spec, estimate effort, judge the quality of the prior work, or recommend porting versus continuing.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mindstudio-ai/remy",
-  "version": "0.1.318",
+  "version": "0.1.320",
   "description": "Remy coding agent",
   "repository": {
     "type": "git",