feat: finish continuous authoring workspace

docs/current_roadmap.md

@@ -173,7 +173,7 @@ Implementation order:
     Astro remains an option, not the default next surface.
 25. Completed: presentation agent authoring story creates a canonical prepared
     authoring recording, Scene 8 as an authentic full-screen conversation,
-    Scene 9 as a 5-phase lifecycle with trace panel, and route-level coverage
+    Scene 9 as a 5-phase lifecycle with synchronized chat dock, and route-level coverage
     that never calls workflow authoring RPC operations during Scene 9
     navigation. Implementation:
     [`presentation agent authoring story`](historical/superpowers/plans/2026-07-11-presentation-agent-authoring-story.md).
@@ -200,8 +200,12 @@ Presentation visual audit, July 10:
 
 Scenes 8 and 9 are now implemented as authentic external-agent handoff and
 factual prepared-authoring proof. The canonical recording, full-screen Scene 8,
-Scene 9 lifecycle with trace panel, and route coverage are all in place. Plan:
+Scene 9 lifecycle with a synchronized chat dock, five factual phase visuals,
-[`presentation agent authoring story`](historical/superpowers/plans/2026-07-11-presentation-agent-authoring-story.md).
+and route coverage are all in place. The obsolete receipt/trace modal was
+removed. Plans:
+[`presentation agent authoring story`](historical/superpowers/plans/2026-07-11-presentation-agent-authoring-story.md)
+and
+[`continuous authoring workspace`](historical/superpowers/plans/2026-07-11-continuous-authoring-workspace.md).
 
 Recommended next visual slices:
 

docs/historical/superpowers/plans/2026-07-11-continuous-authoring-workspace.md

@@ -1,5 +1,7 @@
 # Continuous Authoring Workspace Implementation Plan
 
+Status: Completed on 2026-07-11.
+
 > **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
 
 **Goal:** Rebuild Scenes 8 and 9 as one continuous assistant conversation that synchronizes prepared tool activity with five factual workflow-authoring projections.

web/README.md

@@ -170,8 +170,10 @@ both modes. Replay is visibly labeled and does not create real issues.
 
 The console exposes `/present`, a 720p no-scroll defense compositor for the
 prepared `lda_report_workflow` story. It renders a 14-scene, multi-beat
-storyboard with an adaptive aspect-ratio canvas, stable stage regions, discussion
+storyboard (expanded from the original 12-scene defense plan; items 13 and 14
-branches, act themes, a chat dock, and keyboard navigation.
+now cover evaluation and closing) with an adaptive aspect-ratio canvas, stable
+stage regions, discussion branches, act themes, a chat dock, and keyboard
+navigation.
 
 The final presentation beats frame the evaluation as bounded evidence, make
 claim boundaries and future work explicit, and end on the canonical defense
@@ -180,8 +182,11 @@ discussion index rather than a benchmark or generic conclusion.
 Scenes 8 and 9 use the canonical prepared-authoring recording as their only
 execution evidence. Scene 8 is a full-screen conversation that establishes the
 external agent handoff; Scene 9 breaks the prepared authoring into five phases
-with an overlay trace panel. Neither scene calls workflow authoring RPC
+with the same conversation contracted into a synchronized bottom dock. Each
-operations — they consume deterministic prepared data. Scenes 10 through 12 use
+phase opens its matching prepared tool group beneath a factual source, graph,
+repair, artifact, or deployment view. Neither scene calls workflow authoring
+RPC operations — they consume deterministic prepared data. Scenes 10 through
+12 use
 the canonical replay to show run, interrupt, approval, resume, output, and
 trace. Raw protocol payloads are available through the evidence receipt and
 inspector.
@@ -192,17 +197,20 @@ prepared timeline agent and approval flow; a future AI SDK driver should target
 `AgentMessagePart` / `TimelineAgent`-compatible events instead of replacing the
 presentation timeline.
 
-The key defense states are directly addressable:
+The key deep-link-addressable defense states include:
 
-- `/present#scene/workflow-demo/operation`
-- `/present#scene/workflow-demo/graph`
-- `/present#scene/workflow-demo/interrupt`
-- `/present#scene/interrupt-evidence/approval`
 - `/present#scene/agent-handoff/request` — Scene 8, prepared handoff request
 - `/present#scene/agent-handoff/handoff` — Scene 8, agent accepts
 - `/present#scene/prepared-lifecycle/discover` — Scene 9, discover phase
 - `/present#scene/prepared-lifecycle/draft` — Scene 9, draft phase
 - `/present#scene/prepared-lifecycle/deployment` — Scene 9, deployment phase
+- `/present#scene/run-from-deployment/operation` — Scene 10, run operation
+- `/present#scene/run-from-deployment/graph` — Scene 10, workflow graph
+- `/present#scene/typed-human-boundary/approval` — Scene 11, typed approval
+- `/present#scene/resume-output-evidence/resume` — Scene 12, resume proof
+
+Legacy aliases from the earlier 12-scene plan (such as `workflow-demo` and
+`interrupt-evidence`) are replaced by the IDs above and no longer resolve.
 
 Scene 12 is factual by design. It projects the reviewed/live run into visible
 workflow input, interrupt payload, resume decision, output, and trace facts.
@@ -324,12 +332,13 @@ workflow authoring RPC operations.
 - **Scene 8 (Agent Handoff)**: a full-screen prepared conversation that
   establishes the external agent interface. The `request` beat shows the
   operator requesting a report workflow; the `handoff` beat reveals the full
-  completed conversation across all authoring phases.
+  completed conversation with prepared `wf` tool groups across all authoring
+  phases.
 - **Scene 9 (Prepared Workflow Lifecycle)**: a five-phase lifecycle
   (discover, draft, validate, artifact, deployment) with a compact phase rail
-  and one dominant command/result projection per beat. An overlay
+  and one dominant factual product projection per beat. The Scene 8 thread
-  `AuthoringTracePanel` surfaces the full agent trace for the active phase
+  remains visible as a compact bottom dock; its active tool group follows the
-  without competing with the lifecycle projection.
+  current beat. There is no detached trace modal or second transcript.
 
 The authoring scenes consume deterministic prepared data and never call
 workflow authoring RPCs — only `workflow.health` infrastructure pings are

web/apps/console/src/demo/agent/events.ts

@@ -1,4 +1,6 @@
-import type { AgentToolName } from "./tools.js";
+import type { AgentToolName, PreparedAuthoringToolName } from "./tools.js";
+
+export type AgentMessageToolName = AgentToolName | PreparedAuthoringToolName;
 
 export type AgentRole = "user" | "assistant";
 
@@ -9,13 +11,13 @@ export type PresentationToolAction =
 
 export type AgentToolCall = {
   readonly id: string;
-  readonly name: AgentToolName;
+  readonly name: AgentMessageToolName;
   readonly input: unknown;
 };
 
 export type AgentToolResult = {
   readonly callId: string;
-  readonly name: AgentToolName;
+  readonly name: AgentMessageToolName;
   readonly status: "success" | "failure";
   readonly output: unknown;
 };
@@ -56,7 +58,7 @@ export const agentTextMessage = (id: string, role: AgentRole, text: string): Age
 
 export const agentToolCallPart = (
   id: string,
-  name: AgentToolName,
+  name: AgentMessageToolName,
   input: unknown,
 ): AgentMessagePart => ({
   type: "tool-call",
@@ -65,7 +67,7 @@ export const agentToolCallPart = (
 
 export const agentToolResultPart = (
   callId: string,
-  name: AgentToolName,
+  name: AgentMessageToolName,
   status: "success" | "failure",
   output: unknown,
 ): AgentMessagePart => ({

web/apps/console/src/demo/agent/tools.ts

@@ -5,6 +5,23 @@ export type WorkflowToolName =
   | "readRunTrace"
   | "runWorkflowCommand";
 
+/** Names emitted by the prepared authoring recording for its operation trace. */
+export type PreparedAuthoringToolName =
+  | "workflow.sources.list"
+  | "workflow.capabilities.list"
+  | "workflow.capabilities.inspect"
+  | "wf schema"
+  | "workflow.draft_workspaces.create_from_capability"
+  | "workflow.draft_workspaces.add_step_from_capability"
+  | "workflow.draft_workspaces.get"
+  | "workflow.draft_workspaces.validate"
+  | "workflow.draft_workspaces.set_step_output_map"
+  | "workflow.draft_workspaces.compile"
+  | "workflow.draft_workspaces.create_artifact"
+  | "workflow.artifacts.inspect"
+  | "workflow.deployments.save"
+  | "workflow.deployments.validate";
+
 export type PresentationToolName =
   | "selectWorkflowNode"
   | "focusOperation"

web/apps/console/src/presentation/PresentationRoute.test.tsx

@@ -66,9 +66,9 @@ describe("PresentationRoute", () => {
     window.location.hash = "#scene/agent-handoff/request";
     const { PresentationRoute } = await import("./PresentationRoute.js");
     render(<PresentationRoute />);
-    expect(screen.getByRole("heading", { name: /Agent Handoff/i })).toBeInTheDocument();
+    expect(screen.getByRole("log", { name: "prepared authoring conversation" })).toBeInTheDocument();
     await userEvent.keyboard("{ArrowRight}");
-    expect(await screen.findByText(/The interface delegates durable work to lda\.chat/i)).toBeInTheDocument();
+    expect(await screen.findByRole("button", { name: /deployment.*2 tool calls/i })).toBeInTheDocument();
   });
 
   it("renders audience progress chrome without rail or mode label", async () => {
@@ -373,8 +373,8 @@ describe("PresentationRoute", () => {
     const { PresentationRoute } = await import("./PresentationRoute.js");
     render(<PresentationRoute />);
 
-    expect(await screen.findByRole("heading", { name: "Agent Handoff" })).toBeInTheDocument();
+    expect(await screen.findByRole("log", { name: "prepared authoring conversation" })).toBeInTheDocument();
-    expect(screen.getByText(/A thin agent interface receives the report request/)).toBeInTheDocument();
+    expect(screen.getByRole("button", { name: /discover.*4 tool calls/i })).toBeInTheDocument();
   });
 
   it("navigates to Scene 8 handoff beat via hash", async () => {
@@ -382,8 +382,8 @@ describe("PresentationRoute", () => {
     const { PresentationRoute } = await import("./PresentationRoute.js");
     render(<PresentationRoute />);
 
-    expect(await screen.findByRole("heading", { name: "Agent Handoff" })).toBeInTheDocument();
+    expect(await screen.findByRole("log", { name: "prepared authoring conversation" })).toBeInTheDocument();
-    expect(screen.getByText(/The interface delegates durable work to lda\.chat/)).toBeInTheDocument();
+    expect(screen.getByRole("button", { name: /deployment.*2 tool calls/i })).toBeInTheDocument();
   });
 
   it.each([

web/apps/console/src/presentation/authoring/AgentHandoffScene.test.tsx

@@ -36,7 +36,7 @@ describe("AgentHandoffScene", () => {
 
   it("interleaves prepared workflow tool groups with the handoff conversation", () => {
     renderBeat("handoff");
-    expect(screen.getAllByText(/runWorkflowCommand/i).length).toBeGreaterThan(0);
+    expect(screen.getAllByText(/workflow\.deployments\.save/i).length).toBeGreaterThan(0);
     expect(screen.getByRole("button", { name: /deployment.*2 tool calls/i }))
       .toHaveAttribute("aria-expanded", "true");
   });

web/apps/console/src/presentation/authoring/AuthoringConversation.test.tsx

@@ -16,7 +16,7 @@ describe("AuthoringConversation", () => {
 
     expect(screen.getByRole("log", { name: "prepared authoring conversation" }))
       .toHaveAttribute("data-surface", "stage");
-    expect(screen.getAllByText(/runWorkflowCommand/i).length).toBeGreaterThan(0);
+    expect(screen.getAllByText(/workflow\.deployments\.save/i).length).toBeGreaterThan(0);
     expect(screen.getAllByText(/deployment/i).length).toBeGreaterThan(0);
   });
 

web/apps/console/src/presentation/authoring/AuthoringPhaseVisual.tsx

@@ -15,7 +15,7 @@ type AuthoringPhaseVisualProps = {
 };
 
 const InventoryVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["visual"], { kind: "inventory" }> }) => (
-  <section className="authoring-visual authoring-visual--inventory" role="region" aria-label="discovery evidence">
+  <section className="authoring-visual authoring-visual--inventory" aria-label="discovery evidence">
     <div className="authoring-inventory__sources">
       {visual.sources.map((source) => (
         <div key={source}><Database aria-hidden="true" /><code>{source}</code></div>
@@ -31,7 +31,7 @@ const InventoryVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection[
 );
 
 const GraphVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["visual"], { kind: "graph" }> }) => (
-  <section className="authoring-visual authoring-visual--graph" role="region" aria-label="draft graph evidence">
+  <section className="authoring-visual authoring-visual--graph" aria-label="draft graph evidence">
     <div className="authoring-graph__node"><Database aria-hidden="true" /><strong>{visual.nodes[0]}</strong></div>
     <div className="authoring-graph__edge"><span>{visual.inputBinding}</span><ArrowRight aria-hidden="true" /></div>
     <div className="authoring-graph__node"><Workflow aria-hidden="true" /><strong>{visual.nodes[1]}</strong></div>
@@ -41,7 +41,7 @@ const GraphVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["vis
 );
 
 const RepairVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["visual"], { kind: "repair" }> }) => (
-  <section className="authoring-visual authoring-visual--repair" role="region" aria-label="validation repair evidence">
+  <section className="authoring-visual authoring-visual--repair" aria-label="validation repair evidence">
     <div className="authoring-repair__diagnostic"><AlertTriangle aria-hidden="true" /><span>Diagnostic</span><strong>{visual.diagnostic}</strong></div>
     <ArrowRight aria-hidden="true" />
     <div className="authoring-repair__correction"><Link2 aria-hidden="true" /><span>Repair</span><code>{visual.correction}</code></div>
@@ -50,7 +50,7 @@ const RepairVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["vi
 );
 
 const ArtifactVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["visual"], { kind: "artifact" }> }) => (
-  <section className="authoring-visual authoring-visual--artifact" role="region" aria-label="artifact evidence">
+  <section className="authoring-visual authoring-visual--artifact" aria-label="artifact evidence">
     <LockKeyhole aria-hidden="true" />
     <div><span>Immutable workflow artifact</span><strong>{visual.artifactId}</strong></div>
     <dl><div><dt>Version</dt><dd>Version {visual.version}</dd></div><div><dt>Requirements</dt><dd>{visual.requiredSources} local sources</dd></div></dl>
@@ -58,7 +58,7 @@ const ArtifactVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["
 );
 
 const BindingsVisual = ({ visual }: { visual: Extract<AuthoringPhaseProjection["visual"], { kind: "bindings" }> }) => (
-  <section className="authoring-visual authoring-visual--bindings" role="region" aria-label="deployment binding evidence">
+  <section className="authoring-visual authoring-visual--bindings" aria-label="deployment binding evidence">
     <header><Link2 aria-hidden="true" /><div><span>Deployment</span><strong>{visual.deploymentId}</strong></div><b><CheckCircle2 aria-hidden="true" />{visual.status}</b></header>
     <div className="authoring-bindings__rows">
       {visual.bindings.map((binding) => (

web/apps/console/src/presentation/authoring/authoring-recording.test.ts

@@ -67,6 +67,29 @@ describe("projectPreparedAuthoring", () => {
     }
   });
 
+  it("labels each prepared command with its actual operation name", () => {
+    const titles = projectPreparedAuthoring()
+      .flatMap((phase) => phase.commands)
+      .map((command) => command.title);
+
+    expect(titles).toEqual([
+      "workflow.sources.list",
+      "workflow.capabilities.list",
+      "workflow.capabilities.inspect",
+      "wf schema",
+      "workflow.draft_workspaces.create_from_capability",
+      "workflow.draft_workspaces.add_step_from_capability",
+      "workflow.draft_workspaces.get",
+      "workflow.draft_workspaces.validate",
+      "workflow.draft_workspaces.set_step_output_map",
+      "workflow.draft_workspaces.compile",
+      "workflow.draft_workspaces.create_artifact",
+      "workflow.artifacts.inspect",
+      "workflow.deployments.save",
+      "workflow.deployments.validate",
+    ]);
+  });
+
   it("starts every command with wf or uv run wf", () => {
     const phases = projectPreparedAuthoring();
     for (const phase of phases) {
@@ -118,7 +141,12 @@ describe("projectPreparedAuthoringThread", () => {
 
     expect(calls).toHaveLength(projectPreparedAuthoring()[0]!.commands.length);
     expect(results).toHaveLength(calls.length);
-    expect(calls.every((part) => part.call.name === "runWorkflowCommand")).toBe(true);
+    expect(calls.map((part) => part.call.name)).toEqual([
+      "workflow.sources.list",
+      "workflow.capabilities.list",
+      "workflow.capabilities.inspect",
+      "wf schema",
+    ]);
   });
 
   it("uses a stable phase tool-group id", () => {

web/apps/console/src/presentation/authoring/authoring-recording.ts

@@ -13,6 +13,7 @@ import {
   agentToolResultPart,
   type AgentMessage,
 } from "../../demo/agent/events.js";
+import type { PreparedAuthoringToolName } from "../../demo/agent/tools.js";
 
 export type AuthoringPhaseId =
   | "discover"
@@ -24,6 +25,8 @@ export type AuthoringPhaseId =
 export type CommandResult = "success" | "diagnostic";
 
 export type PreparedAuthoringCommand = {
+  /** JSON-RPC method, or the explicit local CLI label for non-RPC commands. */
+  readonly title: PreparedAuthoringToolName;
   readonly command: string;
   readonly summary: string;
   readonly result: CommandResult;
@@ -70,24 +73,28 @@ const recording: readonly PreparedAuthoringPhase[] = [
     label: "Discover",
     commands: [
       {
+        title: "workflow.sources.list",
         command: "wf source list",
         summary: "List available capability sources",
         result: "success",
         detail: "6 sources: local.lda_docs, local.lda_report, local.issue_board, and platform helpers.",
       },
       {
+        title: "workflow.capabilities.list",
         command: "wf cap list --source local.lda_report --format ids",
         summary: "List report workflow capabilities",
         result: "success",
         detail: "5 capabilities: analyze_documents, build_report, create_issue_drafts, finalise_report, record_revision_request.",
       },
       {
+        title: "workflow.capabilities.inspect",
         command: "wf cap inspect local.lda_report.analyze_documents",
         summary: "Inspect the report-analysis contract",
         result: "success",
         detail: "Input: documents. Output: analysis. Declared outcome: ok.",
       },
       {
+        title: "wf schema",
         command: "wf schema",
         summary: "List registered workflow schemas",
         result: "success",
@@ -116,17 +123,20 @@ const recording: readonly PreparedAuthoringPhase[] = [
     label: "Draft",
     commands: [
       {
+        title: "workflow.draft_workspaces.create_from_capability",
         command: "wf draft create lda_report_workflow --capability local.lda_docs.read_documents",
         summary: "Create a new workflow draft",
         result: "success",
         detail: "Draft 'lda_report_workflow' created with ID draft_a1b2c3",
       },
       {
+        title: "workflow.draft_workspaces.add_step_from_capability",
         command: "wf draft add-step lda_report_workflow --revision 1 --step analyze --capability local.lda_report.analyze_documents --from-step read_documents --from-outcome ok --route ok=__end__ --input state.documents=documents",
         summary: "Add the report-analysis step with a declared route",
         result: "success",
       },
       {
+        title: "workflow.draft_workspaces.get",
         command: "wf draft inspect lda_report_workflow --include-draft",
         summary: "Inspect the prepared draft graph",
         result: "success",
@@ -151,12 +161,14 @@ const recording: readonly PreparedAuthoringPhase[] = [
     label: "Validate",
     commands: [
       {
+        title: "workflow.draft_workspaces.validate",
         command: "wf draft validate lda_report_workflow",
         summary: "Validate the workflow draft",
         result: "diagnostic",
         detail: "Diagnostic: analyze output 'analysis' has no state projection.",
       },
       {
+        title: "workflow.draft_workspaces.set_step_output_map",
         command: "wf draft set-output lda_report_workflow --revision 2 --step analyze --map analysis=state.analysis",
         summary: "Repair the missing output binding",
         result: "success",
@@ -185,18 +197,21 @@ const recording: readonly PreparedAuthoringPhase[] = [
     label: "Artifact",
     commands: [
       {
+        title: "workflow.draft_workspaces.compile",
         command: "wf draft compile lda_report_workflow",
         summary: "Compile the validated draft without mutating it",
         result: "success",
         detail: "Compiled raw plan requires local.lda_docs, local.lda_report, and local.issue_board.",
       },
       {
+        title: "workflow.draft_workspaces.create_artifact",
         command: "wf draft save lda_report_workflow --artifact lda_report_case_study --version 1 --title \"lda.chat Report Case Study\" --outcome completed --outcome cancelled --binding local.lda_docs=local.lda_docs --binding local.lda_report=local.lda_report --binding local.issue_board=local.issue_board",
         summary: "Save immutable artifact version 1",
         result: "success",
         detail: "Artifact lda_report_case_study v1 saved with report-workflow source requirements.",
       },
       {
+        title: "workflow.artifacts.inspect",
         command: "wf artifact inspect lda_report_case_study --version 1",
         summary: "Inspect the compiled artifact",
         result: "success",
@@ -221,12 +236,14 @@ const recording: readonly PreparedAuthoringPhase[] = [
     label: "Deployment",
     commands: [
       {
+        title: "workflow.deployments.save",
         command: "wf deploy save lda_report_case_study.default --artifact lda_report_case_study --version 1 --binding local.lda_docs=local.lda_docs --binding local.lda_report=local.lda_report --binding local.issue_board=local.issue_board",
         summary: "Save deployment with source bindings",
         result: "success",
         detail: "Deployment lda_report_case_study.default saved with all three local source bindings.",
       },
       {
+        title: "workflow.deployments.validate",
         command: "wf deploy validate lda_report_case_study.default",
         summary: "Validate deployment bindings",
         result: "success",
@@ -275,12 +292,12 @@ export const projectPreparedAuthoringThread = (
     const toolParts = phase.commands.flatMap((command, index) => {
       const callId = `${groupId}-command-${index}`;
       return [
-        agentToolCallPart(callId, "runWorkflowCommand", {
+        agentToolCallPart(callId, command.title, {
           phase: phase.phase,
           command: command.command,
           summary: command.summary,
         }),
-        agentToolResultPart(callId, "runWorkflowCommand", "success", {
+        agentToolResultPart(callId, command.title, "success", {
           status: command.result,
           detail: command.detail ?? null,
         }),
@@ -298,27 +315,5 @@ export const projectPreparedAuthoringThread = (
   });
 };
 
-const handoffConversation: readonly AuthoringConversationTurn[] = [
-  {
-    role: "user",
-    text: "We need to prepare a report workflow for the lda_report scenario. What sources and capabilities are available?",
-  },
-  {
-    role: "assistant",
-    text: "I found the local document, report, and issue-board capabilities. I can author the reusable workflow through the public wf CLI.",
-  },
-  {
-    role: "user",
-    text: "Keep the workflow inspectable and validate the source bindings before it runs.",
-  },
-  {
-    role: "assistant",
-    text: "The prepared workflow is ready for deployment: I will create the draft, repair validation diagnostics, save artifact lda_report_case_study v1, then validate its deployment bindings.",
-  },
-];
-
 /** Returns the full prepared authoring recording. */
 export const projectPreparedAuthoring = (): readonly PreparedAuthoringPhase[] => recording;
-
-/** Returns the compact Scene 8 handoff derived from the prepared authoring evidence. */
-export const projectPreparedAuthoringHandoff = (): readonly AuthoringConversationTurn[] => handoffConversation;

web/apps/console/src/presentation/chat/AssistantOperatorThread.tsx

@@ -157,13 +157,12 @@ const AssistantMessageBody = ({
     const part = parts[index]!;
     if (part.type === "text") {
       rendered.push(
-        <p key={`text-${index}`} style={{ whiteSpace: "pre-line" }}>{part.text}</p>,
+        <p key={`text-${part.text}`} style={{ whiteSpace: "pre-line" }}>{part.text}</p>,
       );
       index += 1;
       continue;
     }
 
-    const toolRunStart = index;
     const toolRun: ToolRenderPart[] = [];
     while (index < parts.length && parts[index]!.type !== "text") {
       toolRun.push(parts[index]! as ToolRenderPart);
@@ -176,7 +175,7 @@ const AssistantMessageBody = ({
 
     if (logicalTools.length === 1 && !messageId.startsWith("authoring-")) {
       rendered.push(
-        <div key={`tool-${toolRunStart}`}>
+        <div key={`tool-${logicalTools[0]!.toolCallId ?? logicalTools[0]!.toolName}`}>
           {renderContentPart(logicalTools[0]!, submitApproval, cancelApproval)}
         </div>,
       );
@@ -189,7 +188,7 @@ const AssistantMessageBody = ({
     const count = logicalTools.length;
     rendered.push(
       <ToolGroupRoot
-        key={`tool-group-${toolRunStart}`}
+        key={`tool-group-${groupId}`}
         data-tool-group-id={groupId}
         {...(phaseLabel
           ? {
@@ -203,13 +202,13 @@ const AssistantMessageBody = ({
           {...(phaseLabel ? { label: `${phaseLabel} · ${count} tool ${count === 1 ? "call" : "calls"}` } : {})}
         />
         <ToolGroupContent>
-          {logicalTools.map((tool, toolIndex) => {
+          {logicalTools.map((tool) => {
             const pairedResult = tool.type === "tool-call"
               ? toolRun.find((candidate): candidate is Extract<ToolRenderPart, { readonly type: "tool-result" }> =>
                   candidate.type === "tool-result" && candidate.toolCallId === tool.toolCallId)
               : undefined;
             return (
-            <div key={`${tool.type}-${tool.toolName}-${tool.toolCallId ?? "no-id"}-${toolIndex}`}>
+            <div key={`${tool.type}-${tool.toolName}-${tool.toolCallId ?? "no-id"}`}>
               {renderContentPart(tool, submitApproval, cancelApproval, false, pairedResult)}
             </div>
             );
@@ -254,13 +253,19 @@ export const AssistantOperatorThread = ({
 }: AssistantOperatorThreadProps) => {
   const projected = useMemo(() => projectAgentMessagesForAssistant(messages), [messages]);
   const viewportRef = useRef<HTMLDivElement>(null);
-  const [openToolGroups, setOpenToolGroups] = useState<ReadonlySet<string>>(
+  const [toolGroupOverrides, setToolGroupOverrides] = useState<ReadonlyMap<string, boolean>>(
-    () => new Set(activeToolGroupId ? [activeToolGroupId] : []),
+    () => new Map(),
   );
-
+  const openToolGroups = useMemo(() => {
-  useEffect(() => {
+    const open = new Set<string>();
-    if (activeToolGroupId) setOpenToolGroups(new Set([activeToolGroupId]));
+    if (activeToolGroupId && toolGroupOverrides.get(activeToolGroupId) !== false) {
-  }, [activeToolGroupId]);
+      open.add(activeToolGroupId);
+    }
+    for (const [groupId, isOpen] of toolGroupOverrides) {
+      if (isOpen) open.add(groupId);
+    }
+    return open;
+  }, [activeToolGroupId, toolGroupOverrides]);
 
   useEffect(() => {
     if (!activeToolGroupId) return;
@@ -278,10 +283,9 @@ export const AssistantOperatorThread = ({
   }, [activeToolGroupId, projected]);
 
   const setToolGroupOpen = useCallback((groupId: string, open: boolean) => {
-    setOpenToolGroups((current) => {
+    setToolGroupOverrides((current) => {
-      const next = new Set(current);
+      const next = new Map(current);
-      if (open) next.add(groupId);
+      next.set(groupId, open);
-      else next.delete(groupId);
       return next;
     });
   }, []);
@@ -305,8 +309,8 @@ export const AssistantOperatorThread = ({
             if (message.role === "user") {
               return (
                 <MessageBubble key={message.id} role="user">
-                  {message.content.filter((p) => p.type === "text").map((p, i) => (
+                  {message.content.filter((p) => p.type === "text").map((p) => (
-                    <p key={`text-${i}`} style={{ whiteSpace: "pre-line" }}>{p.text}</p>
+                    <p key={`text-${p.text}`} style={{ whiteSpace: "pre-line" }}>{p.text}</p>
                   ))}
                 </MessageBubble>
               );

web/apps/console/src/presentation/chat/assistantRuntimeProjection.test.ts

@@ -137,4 +137,34 @@ describe("assistantRuntimeProjection", () => {
       { type: "text", text: "provider failed" },
     ]);
   });
+
+  it("assigns sequential indexes for same-type presentation actions", () => {
+    const messages: ReadonlyArray<AgentMessage> = [
+      {
+        id: "dual",
+        role: "assistant",
+        parts: [
+          { type: "presentation-action", action: { type: "selectWorkflowNode", nodeId: "review_issues" } },
+          { type: "presentation-action", action: { type: "selectWorkflowNode", nodeId: "start_step" } },
+        ],
+      },
+    ];
+
+    const projected = projectAgentMessagesForAssistant(messages);
+
+    expect(projected[0]?.content).toEqual([
+      {
+        type: "tool-call",
+        toolCallId: "presentation-dual-0-selectWorkflowNode",
+        toolName: "presentation.selectWorkflowNode",
+        args: { type: "selectWorkflowNode", nodeId: "review_issues" },
+      },
+      {
+        type: "tool-call",
+        toolCallId: "presentation-dual-1-selectWorkflowNode",
+        toolName: "presentation.selectWorkflowNode",
+        args: { type: "selectWorkflowNode", nodeId: "start_step" },
+      },
+    ]);
+  });
 });