🤖 fix: avoid invalid requests from tool-only partials

docs/AGENTS.md

@@ -41,6 +41,7 @@ gh pr view <number> --json mergeable,mergeStateStatus | jq '.'
 ## Repo Reference
 
 - Core files: `src/main.ts`, `src/preload.ts`, `src/App.tsx`, `src/config.ts`.
+- Up-to-date model names: see `src/common/knownModels.ts` for current provider model IDs.
 - Persistent data: `~/.mux/config.json`, `~/.mux/src/<project>/<branch>` (worktrees), `~/.mux/sessions/<workspace>/chat.jsonl`.
 
 ## Documentation Rules
@@ -69,6 +70,12 @@ gh pr view <number> --json mergeable,mergeStateStatus | jq '.'
 - Use `git mv` to retain history when moving files.
 - Never kill the running mux process; rely on `make typecheck` + targeted `bun test path/to/file.test.ts` for validation (run `make test` only when necessary; it can be slow).
 
+## Self-Healing & Crash Resilience
+
+- Prefer **self-healing** behavior: if corrupted or invalid data exists in persisted state (e.g., `chat.jsonl`), the system should sanitize or filter it at load/request time rather than failing permanently.
+- Never let a single malformed line in history brick a workspace—apply defensive filtering in request-building paths so the user can continue working.
+- When streaming crashes, any incomplete state committed to disk should either be repairable on next load or excluded from provider requests to avoid API validation errors.
+
 ## Testing Doctrine
 
 Two types of tests are preferred:

src/browser/stories/App.projectSettings.stories.tsx

@@ -148,7 +148,7 @@ async function openProjectSettings(canvasElement: HTMLElement): Promise<void> {
   const settingsButton = await canvas.findByTestId("settings-button", {}, { timeout: 10000 });
   await userEvent.click(settingsButton);
 
-  await body.findByRole("dialog");
+  await body.findByRole("dialog", {}, { timeout: 10000 });
 
   const projectsButton = await body.findByRole("button", { name: /Projects/i });
   await userEvent.click(projectsButton);
@@ -171,7 +171,7 @@ async function openWorkspaceMCPModal(canvasElement: HTMLElement): Promise<void>
   await userEvent.click(mcpButton);
 
   // Wait for dialog
-  await body.findByRole("dialog");
+  await body.findByRole("dialog", {}, { timeout: 10000 });
 }
 
 // ═══════════════════════════════════════════════════════════════════════════════

src/browser/utils/messages/ChatEventProcessor.ts

@@ -254,12 +254,27 @@ export function createChatEventProcessor(): ChatEventProcessor {
 
       const lastPart = message.parts.at(-1);
       if (lastPart?.type === "reasoning") {
-        lastPart.text += event.delta;
+        // Signature updates come with empty delta - just update the signature
+        if (event.signature && !event.delta) {
+          lastPart.signature = event.signature;
+          lastPart.providerOptions = { anthropic: { signature: event.signature } };
+        } else {
+          lastPart.text += event.delta;
+          // Also capture signature if present with text
+          if (event.signature) {
+            lastPart.signature = event.signature;
+            lastPart.providerOptions = { anthropic: { signature: event.signature } };
+          }
+        }
       } else {
         message.parts.push({
           type: "reasoning",
           text: event.delta,
           timestamp: event.timestamp,
+          signature: event.signature,
+          providerOptions: event.signature
+            ? { anthropic: { signature: event.signature } }
+            : undefined,
         });
       }
       return;

src/browser/utils/messages/modelMessageTransform.test.ts

@@ -152,7 +152,7 @@ describe("modelMessageTransform", () => {
       expect(lastAssistant).toBeTruthy();
       expect(Array.isArray(lastAssistant?.content)).toBe(true);
       if (Array.isArray(lastAssistant?.content)) {
-        expect(lastAssistant.content[0]).toEqual({ type: "reasoning", text: "" });
+        expect(lastAssistant.content[0]).toEqual({ type: "reasoning", text: "..." });
       }
     });
     it("should keep text-only messages unchanged", () => {
@@ -1151,6 +1151,70 @@ describe("filterEmptyAssistantMessages", () => {
     expect(result2[0].id).toBe("assistant-1");
   });
 
+  it("should filter out assistant messages with only incomplete tool calls (input-available)", () => {
+    const messages: MuxMessage[] = [
+      {
+        id: "user-1",
+        role: "user",
+        parts: [{ type: "text", text: "Run a command" }],
+        metadata: { timestamp: 1000 },
+      },
+      {
+        id: "assistant-1",
+        role: "assistant",
+        parts: [
+          {
+            type: "dynamic-tool",
+            state: "input-available",
+            toolCallId: "call-1",
+            toolName: "bash",
+            input: { script: "pwd" },
+          },
+        ],
+        metadata: { timestamp: 2000, partial: true },
+      },
+      {
+        id: "user-2",
+        role: "user",
+        parts: [{ type: "text", text: "Continue" }],
+        metadata: { timestamp: 3000 },
+      },
+    ];
+
+    // Incomplete tool calls are dropped by convertToModelMessages(ignoreIncompleteToolCalls: true),
+    // so we must treat them as empty here to avoid generating an invalid request.
+    const result = filterEmptyAssistantMessages(messages, false);
+    expect(result.map((m) => m.id)).toEqual(["user-1", "user-2"]);
+  });
+
+  it("should preserve assistant messages with completed tool calls (output-available)", () => {
+    const messages: MuxMessage[] = [
+      {
+        id: "user-1",
+        role: "user",
+        parts: [{ type: "text", text: "Run a command" }],
+        metadata: { timestamp: 1000 },
+      },
+      {
+        id: "assistant-1",
+        role: "assistant",
+        parts: [
+          {
+            type: "dynamic-tool",
+            state: "output-available",
+            toolCallId: "call-1",
+            toolName: "bash",
+            input: { script: "pwd" },
+            output: { stdout: "/home/user" },
+          },
+        ],
+        metadata: { timestamp: 2000 },
+      },
+    ];
+
+    const result = filterEmptyAssistantMessages(messages, false);
+    expect(result.map((m) => m.id)).toEqual(["user-1", "assistant-1"]);
+  });
   it("should filter out assistant messages with only empty text regardless of preserveReasoningOnly", () => {
     const messages: MuxMessage[] = [
       {

src/browser/utils/messages/modelMessageTransform.ts

@@ -44,10 +44,37 @@ export function filterEmptyAssistantMessages(
       return false;
     }
 
-    // Keep assistant messages that have at least one text or tool part
-    const hasContent = msg.parts.some(
-      (part) => (part.type === "text" && part.text) || part.type === "dynamic-tool"
-    );
+    // Keep assistant messages that have at least one part that will survive
+    // conversion to provider ModelMessages.
+    //
+    // Important: We call convertToModelMessages(..., { ignoreIncompleteToolCalls: true }).
+    // That means *incomplete* tool calls (state: "input-available") will be dropped.
+    // If we treat them as content here, we can end up sending an assistant message that
+    // becomes empty after conversion, which the AI SDK rejects ("all messages must have
+    // non-empty content...") and can brick a workspace after a crash.
+    const hasContent = msg.parts.some((part) => {
+      if (part.type === "text") {
+        return part.text.length > 0;
+      }
+
+      // Reasoning-only messages are handled below (provider-dependent).
+      if (part.type === "reasoning") {
+        return false;
+      }
+
+      if (part.type === "dynamic-tool") {
+        // Only completed tool calls produce content that can be replayed to the model.
+        return part.state === "output-available";
+      }
+
+      // File/image parts count as content.
+      if (part.type === "file") {
+        return true;
+      }
+
+      // Future-proofing: unknown parts should not brick the request.
+      return true;
+    });
 
     if (hasContent) {
       return true;
@@ -463,6 +490,62 @@ function filterReasoningOnlyMessages(messages: ModelMessage[]): ModelMessage[] {
   });
 }
 
+/**
+ * Strip Anthropic reasoning parts that lack a valid signature.
+ *
+ * Anthropic's Extended Thinking API requires thinking blocks to include a signature
+ * for replay. The Vercel AI SDK's Anthropic provider only sends reasoning parts to
+ * the API if they have providerOptions.anthropic.signature. Reasoning parts we create
+ * (placeholders) or from history (where we didn't capture the signature) will be
+ * silently dropped by the SDK.
+ *
+ * If all parts of an assistant message are unsigned reasoning, the SDK drops them all,
+ * leaving an empty message that Anthropic rejects with:
+ * "all messages must have non-empty content except for the optional final assistant message"
+ *
+ * This function removes unsigned reasoning upfront and filters resulting empty messages.
+ *
+ * NOTE: This is Anthropic-specific. Other providers (e.g., OpenAI) handle reasoning
+ * differently and don't require signatures.
+ */
+function stripUnsignedAnthropicReasoning(messages: ModelMessage[]): ModelMessage[] {
+  const stripped = messages.map((msg) => {
+    if (msg.role !== "assistant") {
+      return msg;
+    }
+
+    const assistantMsg = msg;
+    if (typeof assistantMsg.content === "string") {
+      return msg;
+    }
+
+    // Filter out reasoning parts without anthropic.signature in providerOptions
+    const content = assistantMsg.content.filter((part) => {
+      if (part.type !== "reasoning") {
+        return true;
+      }
+      // Check for anthropic.signature in providerOptions
+      const anthropicMeta = (part.providerOptions as { anthropic?: { signature?: string } })
+        ?.anthropic;
+      return anthropicMeta?.signature != null;
+    });
+
+    const result: typeof assistantMsg = { ...assistantMsg, content };
+    return result;
+  });
+
+  // Filter out messages that became empty after stripping reasoning
+  return stripped.filter((msg) => {
+    if (msg.role !== "assistant") {
+      return true;
+    }
+    if (typeof msg.content === "string") {
+      return msg.content.length > 0;
+    }
+    return msg.content.length > 0;
+  });
+}
+
 /**
  * Coalesce consecutive parts of the same type within each message.
  * Streaming creates many individual text/reasoning parts; merge them for easier debugging.
@@ -512,7 +595,6 @@ function coalesceConsecutiveParts(messages: ModelMessage[]): ModelMessage[] {
     };
   });
 }
-
 /**
  * Merge consecutive user messages with newline separators.
  * When filtering removes assistant messages, we can end up with consecutive user messages.
@@ -610,9 +692,10 @@ function ensureAnthropicThinkingBeforeToolCalls(messages: ModelMessage[]): Model
     }
 
     // Anthropic extended thinking requires tool-use assistant messages to start with a thinking block.
-    // If we still have no reasoning available, insert an empty reasoning part as a minimal placeholder.
+    // If we still have no reasoning available, insert a minimal placeholder reasoning part.
+    // NOTE: The text cannot be empty - Anthropic API rejects empty content.
     if (reasoningParts.length === 0) {
-      reasoningParts = [{ type: "reasoning" as const, text: "" }];
+      reasoningParts = [{ type: "reasoning" as const, text: "..." }];
     }
 
     result.push({
@@ -641,7 +724,7 @@ function ensureAnthropicThinkingBeforeToolCalls(messages: ModelMessage[]): Model
       result[i] = {
         ...assistantMsg,
         content: [
-          { type: "reasoning" as const, text: "" },
+          { type: "reasoning" as const, text: "..." },
           { type: "text" as const, text },
         ],
       };
@@ -658,7 +741,7 @@ function ensureAnthropicThinkingBeforeToolCalls(messages: ModelMessage[]): Model
 
     result[i] = {
       ...assistantMsg,
-      content: [{ type: "reasoning" as const, text: "" }, ...content],
+      content: [{ type: "reasoning" as const, text: "..." }, ...content],
     };
     break;
   }
@@ -703,7 +786,9 @@ export function transformModelMessages(
     // Anthropic: When extended thinking is enabled, preserve reasoning-only messages and ensure
     // tool-call messages start with reasoning. When it's disabled, filter reasoning-only messages.
     if (options?.anthropicThinkingEnabled) {
-      reasoningHandled = ensureAnthropicThinkingBeforeToolCalls(split);
+      // First strip reasoning without signatures (SDK will drop them anyway, causing empty messages)
+      const signedReasoning = stripUnsignedAnthropicReasoning(split);
+      reasoningHandled = ensureAnthropicThinkingBeforeToolCalls(signedReasoning);
     } else {
       reasoningHandled = filterReasoningOnlyMessages(split);
     }

src/cli/run.ts

@@ -195,6 +195,7 @@ program
   .option("--json", "output NDJSON for programmatic consumption")
   .option("-q, --quiet", "only output final result")
   .option("--workspace-id <id>", "explicit workspace ID (auto-generated if not provided)")
+  .option("--workspace <id>", "continue an existing workspace (loads history, skips init)")
   .option("--config-root <path>", "mux config directory")
   .option("--mcp <server>", "MCP server as name=command (can be repeated)", collectMcpServers, [])
   .option("--no-mcp-config", "ignore .mux/mcp.jsonc, use only --mcp servers")
@@ -227,6 +228,7 @@ interface CLIOptions {
   json?: boolean;
   quiet?: boolean;
   workspaceId?: string;
+  workspace?: string;
   configRoot?: string;
   mcp: MCPServerEntry[];
   mcpConfig: boolean;
@@ -250,10 +252,6 @@ async function main(): Promise<void> {
   }
   // Default is already "warn" for CLI mode (set in log.ts)
 
-  // Resolve directory
-  const projectDir = path.resolve(opts.dir);
-  await ensureDirectory(projectDir);
-
   // Get message from arg or stdin
   const stdinMessage = await gatherMessageFromStdin();
   const message = messageArg?.trim() ?? stdinMessage.trim();
@@ -266,7 +264,35 @@ async function main(): Promise<void> {
 
   // Setup config
   const config = new Config(opts.configRoot);
-  const workspaceId = opts.workspaceId ?? generateWorkspaceId();
+
+  // Determine if continuing an existing workspace
+  const continueWorkspace = opts.workspace;
+  const workspaceId = continueWorkspace ?? opts.workspaceId ?? generateWorkspaceId();
+
+  // Resolve directory - for continuing workspace, try to get from metadata
+  let projectDir: string;
+  if (continueWorkspace) {
+    const metadataPath = path.join(config.sessionsDir, continueWorkspace, "metadata.json");
+    try {
+      const metadataContent = await fs.readFile(metadataPath, "utf-8");
+      const metadata = JSON.parse(metadataContent) as { projectPath?: string };
+      if (metadata.projectPath) {
+        projectDir = metadata.projectPath;
+        log.info(`Continuing workspace ${continueWorkspace}, using project path: ${projectDir}`);
+      } else {
+        projectDir = path.resolve(opts.dir);
+        log.warn(`No projectPath in metadata, using --dir: ${projectDir}`);
+      }
+    } catch {
+      // Metadata doesn't exist or is invalid, fall back to --dir
+      projectDir = path.resolve(opts.dir);
+      log.warn(`Could not read metadata for ${continueWorkspace}, using --dir: ${projectDir}`);
+    }
+  } else {
+    projectDir = path.resolve(opts.dir);
+    await ensureDirectory(projectDir);
+  }
+
   const model: string = opts.model;
   const runtimeConfig = parseRuntimeConfig(opts.runtime, config.srcDir);
   const thinkingLevel = parseThinkingLevel(opts.thinking);
@@ -333,11 +359,17 @@ async function main(): Promise<void> {
     backgroundProcessManager,
   });
 
-  await session.ensureMetadata({
-    workspacePath: projectDir,
-    projectName: path.basename(projectDir),
-    runtimeConfig,
-  });
+  // For continuing workspace, metadata should already exist
+  // For new workspace, create it
+  if (!continueWorkspace) {
+    await session.ensureMetadata({
+      workspacePath: projectDir,
+      projectName: path.basename(projectDir),
+      runtimeConfig,
+    });
+  } else {
+    log.info(`Continuing workspace ${workspaceId} - using existing metadata`);
+  }
 
   const buildSendOptions = (cliMode: CLIMode): SendMessageOptions => ({
     model,