Integrate OpenHandoff factory workspace (#212)

factory/packages/client/test/e2e/full-integration-e2e.test.ts

@@ -0,0 +1,197 @@
+import { randomUUID } from "node:crypto";
+import { describe, expect, it } from "vitest";
+import type { HistoryEvent, RepoOverview } from "@openhandoff/shared";
+import { createBackendClient } from "../../src/backend-client.js";
+
+const RUN_FULL_E2E = process.env.HF_ENABLE_DAEMON_FULL_E2E === "1";
+
+function requiredEnv(name: string): string {
+  const value = process.env[name]?.trim();
+  if (!value) {
+    throw new Error(`Missing required env var: ${name}`);
+  }
+  return value;
+}
+
+function parseGithubRepo(input: string): { fullName: string } {
+  const trimmed = input.trim();
+  const shorthand = trimmed.match(/^([a-zA-Z0-9_.-]+)\/([a-zA-Z0-9_.-]+)$/);
+  if (shorthand) {
+    return { fullName: `${shorthand[1]}/${shorthand[2]}` };
+  }
+
+  const url = new URL(trimmed.startsWith("http") ? trimmed : `https://${trimmed}`);
+  const parts = url.pathname.replace(/^\/+/, "").split("/").filter(Boolean);
+  if (url.hostname.toLowerCase().includes("github.com") && parts.length >= 2) {
+    return { fullName: `${parts[0]}/${(parts[1] ?? "").replace(/\.git$/, "")}` };
+  }
+
+  throw new Error(`Unable to parse GitHub repo from: ${input}`);
+}
+
+async function sleep(ms: number): Promise<void> {
+  await new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+async function poll<T>(
+  label: string,
+  timeoutMs: number,
+  intervalMs: number,
+  fn: () => Promise<T>,
+  isDone: (value: T) => boolean
+): Promise<T> {
+  const start = Date.now();
+  let last: T;
+  for (;;) {
+    last = await fn();
+    if (isDone(last)) {
+      return last;
+    }
+    if (Date.now() - start > timeoutMs) {
+      throw new Error(`timed out waiting for ${label}`);
+    }
+    await sleep(intervalMs);
+  }
+}
+
+function parseHistoryPayload(event: HistoryEvent): Record<string, unknown> {
+  try {
+    return JSON.parse(event.payloadJson) as Record<string, unknown>;
+  } catch {
+    return {};
+  }
+}
+
+async function githubApi(token: string, path: string, init?: RequestInit): Promise<Response> {
+  const url = `https://api.github.com/${path.replace(/^\/+/, "")}`;
+  return await fetch(url, {
+    ...init,
+    headers: {
+      Accept: "application/vnd.github+json",
+      Authorization: `Bearer ${token}`,
+      "X-GitHub-Api-Version": "2022-11-28",
+      ...(init?.headers ?? {}),
+    },
+  });
+}
+
+async function ensureRemoteBranchExists(
+  token: string,
+  fullName: string,
+  branchName: string
+): Promise<void> {
+  const repoRes = await githubApi(token, `repos/${fullName}`, { method: "GET" });
+  if (!repoRes.ok) {
+    throw new Error(`GitHub repo lookup failed: ${repoRes.status} ${await repoRes.text()}`);
+  }
+  const repo = (await repoRes.json()) as { default_branch?: string };
+  const defaultBranch = repo.default_branch;
+  if (!defaultBranch) {
+    throw new Error(`GitHub repo default branch is missing for ${fullName}`);
+  }
+
+  const defaultRefRes = await githubApi(
+    token,
+    `repos/${fullName}/git/ref/heads/${encodeURIComponent(defaultBranch)}`,
+    { method: "GET" }
+  );
+  if (!defaultRefRes.ok) {
+    throw new Error(`GitHub default ref lookup failed: ${defaultRefRes.status} ${await defaultRefRes.text()}`);
+  }
+  const defaultRef = (await defaultRefRes.json()) as { object?: { sha?: string } };
+  const sha = defaultRef.object?.sha;
+  if (!sha) {
+    throw new Error(`GitHub default ref sha missing for ${fullName}:${defaultBranch}`);
+  }
+
+  const createRefRes = await githubApi(token, `repos/${fullName}/git/refs`, {
+    method: "POST",
+    body: JSON.stringify({
+      ref: `refs/heads/${branchName}`,
+      sha,
+    }),
+    headers: { "Content-Type": "application/json" },
+  });
+  if (createRefRes.ok || createRefRes.status === 422) {
+    return;
+  }
+
+  throw new Error(`GitHub create ref failed: ${createRefRes.status} ${await createRefRes.text()}`);
+}
+
+describe("e2e(client): full integration stack workflow", () => {
+  it.skipIf(!RUN_FULL_E2E)(
+    "adds repo, loads branch graph, and executes a stack restack action",
+    { timeout: 8 * 60_000 },
+    async () => {
+      const endpoint =
+        process.env.HF_E2E_BACKEND_ENDPOINT?.trim() || "http://127.0.0.1:7741/api/rivet";
+      const workspaceId = process.env.HF_E2E_WORKSPACE?.trim() || "default";
+      const repoRemote = requiredEnv("HF_E2E_GITHUB_REPO");
+      const githubToken = requiredEnv("GITHUB_TOKEN");
+      const { fullName } = parseGithubRepo(repoRemote);
+      const normalizedRepoRemote = `https://github.com/${fullName}.git`;
+      const seededBranch = `e2e/full-seed-${Date.now().toString(36)}-${randomUUID().slice(0, 8)}`;
+
+      const client = createBackendClient({
+        endpoint,
+        defaultWorkspaceId: workspaceId,
+      });
+
+      try {
+        await ensureRemoteBranchExists(githubToken, fullName, seededBranch);
+
+        const repo = await client.addRepo(workspaceId, repoRemote);
+        expect(repo.remoteUrl).toBe(normalizedRepoRemote);
+
+        const overview = await poll<RepoOverview>(
+          "repo overview includes seeded branch",
+          90_000,
+          1_000,
+          async () => client.getRepoOverview(workspaceId, repo.repoId),
+          (value) => value.branches.some((row) => row.branchName === seededBranch)
+        );
+
+        if (!overview.stackAvailable) {
+          throw new Error(
+            "git-spice is unavailable for this repo during full integration e2e; set HF_GIT_SPICE_BIN or install git-spice in the backend container"
+          );
+        }
+
+        const stackResult = await client.runRepoStackAction({
+          workspaceId,
+          repoId: repo.repoId,
+          action: "restack_repo",
+        });
+        expect(stackResult.executed).toBe(true);
+        expect(stackResult.action).toBe("restack_repo");
+
+        await poll<HistoryEvent[]>(
+          "repo stack action history event",
+          60_000,
+          1_000,
+          async () => client.listHistory({ workspaceId, limit: 200 }),
+          (events) =>
+            events.some((event) => {
+              if (event.kind !== "repo.stack_action") {
+                return false;
+              }
+              const payload = parseHistoryPayload(event);
+              return payload.action === "restack_repo";
+            })
+        );
+
+        const postActionOverview = await client.getRepoOverview(workspaceId, repo.repoId);
+        const seededRow = postActionOverview.branches.find((row) => row.branchName === seededBranch);
+        expect(Boolean(seededRow)).toBe(true);
+        expect(postActionOverview.fetchedAt).toBeGreaterThan(overview.fetchedAt);
+      } finally {
+        await githubApi(
+          githubToken,
+          `repos/${fullName}/git/refs/heads/${encodeURIComponent(seededBranch)}`,
+          { method: "DELETE" }
+        ).catch(() => {});
+      }
+    }
+  );
+});

factory/packages/client/test/e2e/github-pr-e2e.test.ts

@@ -0,0 +1,351 @@
+import { describe, expect, it } from "vitest";
+import type { HandoffRecord, HistoryEvent } from "@openhandoff/shared";
+import { createBackendClient } from "../../src/backend-client.js";
+
+const RUN_E2E = process.env.HF_ENABLE_DAEMON_E2E === "1";
+
+function requiredEnv(name: string): string {
+  const value = process.env[name]?.trim();
+  if (!value) {
+    throw new Error(`Missing required env var: ${name}`);
+  }
+  return value;
+}
+
+function parseGithubRepo(input: string): { owner: string; repo: string; fullName: string } {
+  const trimmed = input.trim();
+  if (!trimmed) {
+    throw new Error("HF_E2E_GITHUB_REPO is empty");
+  }
+
+  // owner/repo shorthand
+  const shorthand = trimmed.match(/^([a-zA-Z0-9_.-]+)\/([a-zA-Z0-9_.-]+)$/);
+  if (shorthand) {
+    const owner = shorthand[1]!;
+    const repo = shorthand[2]!;
+    return { owner, repo, fullName: `${owner}/${repo}` };
+  }
+
+  // https://github.com/owner/repo(.git)?(/...)?
+  try {
+    const url = new URL(trimmed.startsWith("http") ? trimmed : `https://${trimmed}`);
+    const parts = url.pathname.replace(/^\/+/, "").split("/").filter(Boolean);
+    if (url.hostname.toLowerCase().includes("github.com") && parts.length >= 2) {
+      const owner = parts[0]!;
+      const repo = (parts[1] ?? "").replace(/\.git$/, "");
+      if (owner && repo) {
+        return { owner, repo, fullName: `${owner}/${repo}` };
+      }
+    }
+  } catch {
+    // fall through
+  }
+
+  throw new Error(`Unable to parse GitHub repo from: ${input}`);
+}
+
+async function sleep(ms: number): Promise<void> {
+  await new Promise((r) => setTimeout(r, ms));
+}
+
+async function poll<T>(
+  label: string,
+  timeoutMs: number,
+  intervalMs: number,
+  fn: () => Promise<T>,
+  isDone: (value: T) => boolean,
+  onTick?: (value: T) => void
+): Promise<T> {
+  const start = Date.now();
+  let last: T;
+  for (;;) {
+    last = await fn();
+    onTick?.(last);
+    if (isDone(last)) {
+      return last;
+    }
+    if (Date.now() - start > timeoutMs) {
+      throw new Error(`timed out waiting for ${label}`);
+    }
+    await sleep(intervalMs);
+  }
+}
+
+function parseHistoryPayload(event: HistoryEvent): Record<string, unknown> {
+  try {
+    return JSON.parse(event.payloadJson) as Record<string, unknown>;
+  } catch {
+    return {};
+  }
+}
+
+async function debugDump(client: ReturnType<typeof createBackendClient>, workspaceId: string, handoffId: string): Promise<string> {
+  try {
+    const handoff = await client.getHandoff(workspaceId, handoffId);
+    const history = await client.listHistory({ workspaceId, handoffId, limit: 80 }).catch(() => []);
+    const historySummary = history
+      .slice(0, 20)
+      .map((e) => `${new Date(e.createdAt).toISOString()} ${e.kind}`)
+      .join("\n");
+
+    let sessionEventsSummary = "";
+    if (handoff.activeSandboxId && handoff.activeSessionId) {
+      const events = await client
+        .listSandboxSessionEvents(workspaceId, handoff.providerId, handoff.activeSandboxId, {
+          sessionId: handoff.activeSessionId,
+          limit: 50,
+        })
+        .then((r) => r.items)
+        .catch(() => []);
+      sessionEventsSummary = events
+        .slice(-12)
+        .map((e) => `${new Date(e.createdAt).toISOString()} ${e.sender}`)
+        .join("\n");
+    }
+
+    return [
+      "=== handoff ===",
+      JSON.stringify(
+        {
+          status: handoff.status,
+          statusMessage: handoff.statusMessage,
+          title: handoff.title,
+          branchName: handoff.branchName,
+          activeSandboxId: handoff.activeSandboxId,
+          activeSessionId: handoff.activeSessionId,
+          prUrl: handoff.prUrl,
+          prSubmitted: handoff.prSubmitted,
+        },
+        null,
+        2
+      ),
+      "=== history (most recent first) ===",
+      historySummary || "(none)",
+      "=== session events (tail) ===",
+      sessionEventsSummary || "(none)",
+    ].join("\n");
+  } catch (err) {
+    return `debug dump failed: ${err instanceof Error ? err.message : String(err)}`;
+  }
+}
+
+async function githubApi(token: string, path: string, init?: RequestInit): Promise<Response> {
+  const url = `https://api.github.com/${path.replace(/^\/+/, "")}`;
+  return await fetch(url, {
+    ...init,
+    headers: {
+      Accept: "application/vnd.github+json",
+      Authorization: `Bearer ${token}`,
+      "X-GitHub-Api-Version": "2022-11-28",
+      ...(init?.headers ?? {}),
+    },
+  });
+}
+
+describe("e2e: backend -> sandbox-agent -> git -> PR", () => {
+  it.skipIf(!RUN_E2E)(
+    "creates a handoff, waits for agent to implement, and opens a PR",
+    { timeout: 15 * 60_000 },
+    async () => {
+      const endpoint =
+        process.env.HF_E2E_BACKEND_ENDPOINT?.trim() || "http://127.0.0.1:7741/api/rivet";
+      const workspaceId = process.env.HF_E2E_WORKSPACE?.trim() || "default";
+      const repoRemote = requiredEnv("HF_E2E_GITHUB_REPO");
+      const githubToken = requiredEnv("GITHUB_TOKEN");
+
+      const { fullName } = parseGithubRepo(repoRemote);
+      const runId = `${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 8)}`;
+      const expectedFile = `e2e/${runId}.txt`;
+
+      const client = createBackendClient({
+        endpoint,
+        defaultWorkspaceId: workspaceId,
+      });
+
+      const repo = await client.addRepo(workspaceId, repoRemote);
+
+      const created = await client.createHandoff({
+        workspaceId,
+        repoId: repo.repoId,
+        task: [
+          "E2E test task:",
+          `1. Create a new file at ${expectedFile} containing the single line: ${runId}`,
+          "2. git add the file",
+          `3. git commit -m \"test(e2e): ${runId}\"`,
+          "4. git push the branch to origin",
+          "5. Stop when done (agent should go idle).",
+        ].join("\n"),
+        providerId: "daytona",
+        explicitTitle: `test(e2e): ${runId}`,
+        explicitBranchName: `e2e/${runId}`,
+      });
+
+      let prNumber: number | null = null;
+      let branchName: string | null = null;
+      let sandboxId: string | null = null;
+      let sessionId: string | null = null;
+      let lastStatus: string | null = null;
+
+      try {
+        const namedAndProvisioned = await poll<HandoffRecord>(
+          "handoff naming + sandbox provisioning",
+          // Cold Daytona snapshot/image preparation can exceed 5 minutes on first run.
+          8 * 60_000,
+          1_000,
+          async () => client.getHandoff(workspaceId, created.handoffId),
+          (h) => Boolean(h.title && h.branchName && h.activeSandboxId),
+          (h) => {
+            if (h.status !== lastStatus) {
+              lastStatus = h.status;
+            }
+            if (h.status === "error") {
+              throw new Error("handoff entered error state during provisioning");
+            }
+          }
+        ).catch(async (err) => {
+          const dump = await debugDump(client, workspaceId, created.handoffId);
+          throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+        });
+
+        branchName = namedAndProvisioned.branchName!;
+        sandboxId = namedAndProvisioned.activeSandboxId!;
+
+        const withSession = await poll<HandoffRecord>(
+          "handoff to create active session",
+          3 * 60_000,
+          1_500,
+          async () => client.getHandoff(workspaceId, created.handoffId),
+          (h) => Boolean(h.activeSessionId),
+          (h) => {
+            if (h.status === "error") {
+              throw new Error("handoff entered error state while waiting for active session");
+            }
+          }
+        ).catch(async (err) => {
+          const dump = await debugDump(client, workspaceId, created.handoffId);
+          throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+        });
+
+        sessionId = withSession.activeSessionId!;
+
+        await poll<{ id: string }[]>(
+          "session transcript bootstrap events",
+          2 * 60_000,
+          2_000,
+          async () =>
+            (
+              await client.listSandboxSessionEvents(workspaceId, withSession.providerId, sandboxId!, {
+                sessionId: sessionId!,
+                limit: 40,
+              })
+            ).items,
+          (events) => events.length > 0
+        ).catch(async (err) => {
+          const dump = await debugDump(client, workspaceId, created.handoffId);
+          throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+        });
+
+        await poll<HandoffRecord>(
+          "handoff to reach idle state",
+          8 * 60_000,
+          2_000,
+          async () => client.getHandoff(workspaceId, created.handoffId),
+          (h) => h.status === "idle",
+          (h) => {
+            if (h.status === "error") {
+              throw new Error("handoff entered error state while waiting for idle");
+            }
+          }
+        ).catch(async (err) => {
+          const dump = await debugDump(client, workspaceId, created.handoffId);
+          throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+        });
+
+        const prCreatedEvent = await poll<HistoryEvent[]>(
+          "PR creation history event",
+          3 * 60_000,
+          2_000,
+          async () => client.listHistory({ workspaceId, handoffId: created.handoffId, limit: 200 }),
+          (events) => events.some((e) => e.kind === "handoff.pr_created")
+        )
+          .catch(async (err) => {
+            const dump = await debugDump(client, workspaceId, created.handoffId);
+            throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+          })
+          .then((events) => events.find((e) => e.kind === "handoff.pr_created")!);
+
+        const payload = parseHistoryPayload(prCreatedEvent);
+        prNumber = Number(payload.prNumber);
+        const prUrl = String(payload.prUrl ?? "");
+
+        expect(prNumber).toBeGreaterThan(0);
+        expect(prUrl).toContain("/pull/");
+
+        const prFilesRes = await githubApi(
+          githubToken,
+          `repos/${fullName}/pulls/${prNumber}/files?per_page=100`,
+          { method: "GET" }
+        );
+        if (!prFilesRes.ok) {
+          const body = await prFilesRes.text();
+          throw new Error(`GitHub PR files request failed: ${prFilesRes.status} ${body}`);
+        }
+        const prFiles = (await prFilesRes.json()) as Array<{ filename: string }>;
+        expect(prFiles.some((f) => f.filename === expectedFile)).toBe(true);
+
+        // Close the handoff and assert the sandbox is released (stopped).
+        await client.runAction(workspaceId, created.handoffId, "archive");
+
+        await poll<HandoffRecord>(
+          "handoff to become archived (session released)",
+          60_000,
+          1_000,
+          async () => client.getHandoff(workspaceId, created.handoffId),
+          (h) => h.status === "archived" && h.activeSessionId === null
+        ).catch(async (err) => {
+          const dump = await debugDump(client, workspaceId, created.handoffId);
+          throw new Error(`${err instanceof Error ? err.message : String(err)}\n${dump}`);
+        });
+
+        if (sandboxId) {
+          await poll<{ providerId: string; sandboxId: string; state: string; at: number }>(
+            "daytona sandbox to stop",
+            2 * 60_000,
+            2_000,
+            async () => client.sandboxProviderState(workspaceId, "daytona", sandboxId!),
+            (s) => {
+              const st = String(s.state).toLowerCase();
+              return st.includes("stopped") || st.includes("suspended") || st.includes("paused");
+            }
+          ).catch(async (err) => {
+            const dump = await debugDump(client, workspaceId, created.handoffId);
+            const state = await client
+              .sandboxProviderState(workspaceId, "daytona", sandboxId!)
+              .catch(() => null);
+            throw new Error(
+              `${err instanceof Error ? err.message : String(err)}\n` +
+                `sandbox state: ${state ? state.state : "unknown"}\n` +
+                `${dump}`
+            );
+          });
+        }
+      } finally {
+        if (prNumber && Number.isFinite(prNumber)) {
+          await githubApi(githubToken, `repos/${fullName}/pulls/${prNumber}`, {
+            method: "PATCH",
+            body: JSON.stringify({ state: "closed" }),
+            headers: { "Content-Type": "application/json" },
+          }).catch(() => {});
+        }
+
+        if (branchName) {
+          await githubApi(
+            githubToken,
+            `repos/${fullName}/git/refs/heads/${encodeURIComponent(branchName)}`,
+            { method: "DELETE" }
+          ).catch(() => {});
+        }
+      }
+    }
+  );
+});

factory/packages/client/test/e2e/workbench-e2e.test.ts

@@ -0,0 +1,334 @@
+import { execFile } from "node:child_process";
+import { promisify } from "node:util";
+import { describe, expect, it } from "vitest";
+import type {
+  HandoffWorkbenchSnapshot,
+  WorkbenchAgentTab,
+  WorkbenchHandoff,
+  WorkbenchModelId,
+  WorkbenchTranscriptEvent,
+} from "@openhandoff/shared";
+import { createBackendClient } from "../../src/backend-client.js";
+
+const RUN_WORKBENCH_E2E = process.env.HF_ENABLE_DAEMON_WORKBENCH_E2E === "1";
+const execFileAsync = promisify(execFile);
+
+function requiredEnv(name: string): string {
+  const value = process.env[name]?.trim();
+  if (!value) {
+    throw new Error(`Missing required env var: ${name}`);
+  }
+  return value;
+}
+
+function workbenchModelEnv(name: string, fallback: WorkbenchModelId): WorkbenchModelId {
+  const value = process.env[name]?.trim();
+  switch (value) {
+    case "claude-sonnet-4":
+    case "claude-opus-4":
+    case "gpt-4o":
+    case "o3":
+      return value;
+    default:
+      return fallback;
+  }
+}
+
+async function sleep(ms: number): Promise<void> {
+  await new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+async function seedSandboxFile(workspaceId: string, handoffId: string, filePath: string, content: string): Promise<void> {
+  const repoPath = `/root/.local/share/openhandoff/local-sandboxes/${workspaceId}/${handoffId}/repo`;
+  const script = [
+    `cd ${JSON.stringify(repoPath)}`,
+    `mkdir -p ${JSON.stringify(filePath.includes("/") ? filePath.slice(0, filePath.lastIndexOf("/")) : ".")}`,
+    `printf '%s\\n' ${JSON.stringify(content)} > ${JSON.stringify(filePath)}`,
+  ].join(" && ");
+  await execFileAsync("docker", ["exec", "openhandoff-backend-1", "bash", "-lc", script]);
+}
+
+async function poll<T>(
+  label: string,
+  timeoutMs: number,
+  intervalMs: number,
+  fn: () => Promise<T>,
+  isDone: (value: T) => boolean,
+): Promise<T> {
+  const startedAt = Date.now();
+  let lastValue: T;
+
+  for (;;) {
+    lastValue = await fn();
+    if (isDone(lastValue)) {
+      return lastValue;
+    }
+    if (Date.now() - startedAt > timeoutMs) {
+      throw new Error(`timed out waiting for ${label}`);
+    }
+    await sleep(intervalMs);
+  }
+}
+
+function findHandoff(snapshot: HandoffWorkbenchSnapshot, handoffId: string): WorkbenchHandoff {
+  const handoff = snapshot.handoffs.find((candidate) => candidate.id === handoffId);
+  if (!handoff) {
+    throw new Error(`handoff ${handoffId} missing from snapshot`);
+  }
+  return handoff;
+}
+
+function findTab(handoff: WorkbenchHandoff, tabId: string): WorkbenchAgentTab {
+  const tab = handoff.tabs.find((candidate) => candidate.id === tabId);
+  if (!tab) {
+    throw new Error(`tab ${tabId} missing from handoff ${handoff.id}`);
+  }
+  return tab;
+}
+
+function extractEventText(event: WorkbenchTranscriptEvent): string {
+  const payload = event.payload;
+  if (!payload || typeof payload !== "object") {
+    return String(payload ?? "");
+  }
+
+  const envelope = payload as {
+    method?: unknown;
+    params?: unknown;
+    result?: unknown;
+    error?: unknown;
+  };
+
+  const params = envelope.params;
+  if (params && typeof params === "object") {
+    const update = (params as { update?: unknown }).update;
+    if (update && typeof update === "object") {
+      const content = (update as { content?: unknown }).content;
+      if (content && typeof content === "object") {
+        const chunkText = (content as { text?: unknown }).text;
+        if (typeof chunkText === "string") {
+          return chunkText;
+        }
+      }
+    }
+
+    const text = (params as { text?: unknown }).text;
+    if (typeof text === "string" && text.trim()) {
+      return text.trim();
+    }
+    const prompt = (params as { prompt?: Array<{ text?: unknown }> }).prompt;
+    if (Array.isArray(prompt)) {
+      const value = prompt
+        .map((item) => (typeof item?.text === "string" ? item.text.trim() : ""))
+        .filter(Boolean)
+        .join("\n");
+      if (value) {
+        return value;
+      }
+    }
+  }
+
+  const result = envelope.result;
+  if (result && typeof result === "object") {
+    const text = (result as { text?: unknown }).text;
+    if (typeof text === "string" && text.trim()) {
+      return text.trim();
+    }
+  }
+
+  if (envelope.error) {
+    return JSON.stringify(envelope.error);
+  }
+
+  if (typeof envelope.method === "string") {
+    return envelope.method;
+  }
+
+  return JSON.stringify(payload);
+}
+
+function transcriptIncludesAgentText(
+  transcript: WorkbenchTranscriptEvent[],
+  expectedText: string,
+): boolean {
+  return transcript
+    .filter((event) => event.sender === "agent")
+    .map((event) => extractEventText(event))
+    .join("")
+    .includes(expectedText);
+}
+
+describe("e2e(client): workbench flows", () => {
+  it.skipIf(!RUN_WORKBENCH_E2E)(
+    "creates a handoff, adds sessions, exchanges messages, and manages workbench state",
+    { timeout: 20 * 60_000 },
+    async () => {
+      const endpoint =
+        process.env.HF_E2E_BACKEND_ENDPOINT?.trim() || "http://127.0.0.1:7741/api/rivet";
+      const workspaceId = process.env.HF_E2E_WORKSPACE?.trim() || "default";
+      const repoRemote = requiredEnv("HF_E2E_GITHUB_REPO");
+      const model = workbenchModelEnv("HF_E2E_MODEL", "gpt-4o");
+      const runId = `wb-${Date.now().toString(36)}`;
+      const expectedFile = `${runId}.txt`;
+      const expectedInitialReply = `WORKBENCH_READY_${runId}`;
+      const expectedReply = `WORKBENCH_ACK_${runId}`;
+
+      const client = createBackendClient({
+        endpoint,
+        defaultWorkspaceId: workspaceId,
+      });
+
+      const repo = await client.addRepo(workspaceId, repoRemote);
+      const created = await client.createWorkbenchHandoff(workspaceId, {
+        repoId: repo.repoId,
+        title: `Workbench E2E ${runId}`,
+        branch: `e2e/${runId}`,
+        model,
+        task: `Reply with exactly: ${expectedInitialReply}`,
+      });
+
+      const provisioned = await poll(
+        "handoff provisioning",
+        12 * 60_000,
+        2_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => handoff.branch === `e2e/${runId}` && handoff.tabs.length > 0,
+      );
+
+      const primaryTab = provisioned.tabs[0]!;
+
+      const initialCompleted = await poll(
+        "initial agent response",
+        12 * 60_000,
+        2_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => {
+          const tab = findTab(handoff, primaryTab.id);
+          return (
+            handoff.status === "idle" &&
+            tab.status === "idle" &&
+            transcriptIncludesAgentText(tab.transcript, expectedInitialReply)
+          );
+        },
+      );
+
+      expect(findTab(initialCompleted, primaryTab.id).sessionId).toBeTruthy();
+      expect(transcriptIncludesAgentText(findTab(initialCompleted, primaryTab.id).transcript, expectedInitialReply)).toBe(true);
+
+      await seedSandboxFile(workspaceId, created.handoffId, expectedFile, runId);
+
+      const fileSeeded = await poll(
+        "seeded sandbox file reflected in workbench",
+        30_000,
+        1_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => handoff.fileChanges.some((file) => file.path === expectedFile),
+      );
+      expect(fileSeeded.fileChanges.some((file) => file.path === expectedFile)).toBe(true);
+
+      await client.renameWorkbenchHandoff(workspaceId, {
+        handoffId: created.handoffId,
+        value: `Workbench E2E ${runId} Renamed`,
+      });
+      await client.renameWorkbenchSession(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: primaryTab.id,
+        title: "Primary Session",
+      });
+
+      const secondTab = await client.createWorkbenchSession(workspaceId, {
+        handoffId: created.handoffId,
+        model,
+      });
+
+      await client.renameWorkbenchSession(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: secondTab.tabId,
+        title: "Follow-up Session",
+      });
+
+      await client.updateWorkbenchDraft(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: secondTab.tabId,
+        text: `Reply with exactly: ${expectedReply}`,
+        attachments: [
+          {
+            id: `${expectedFile}:1`,
+            filePath: expectedFile,
+            lineNumber: 1,
+            lineContent: runId,
+          },
+        ],
+      });
+
+      const drafted = findHandoff(await client.getWorkbench(workspaceId), created.handoffId);
+      expect(findTab(drafted, secondTab.tabId).draft.text).toContain(expectedReply);
+      expect(findTab(drafted, secondTab.tabId).draft.attachments).toHaveLength(1);
+
+      await client.sendWorkbenchMessage(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: secondTab.tabId,
+        text: `Reply with exactly: ${expectedReply}`,
+        attachments: [],
+      });
+
+      const withSecondReply = await poll(
+        "follow-up session response",
+        10 * 60_000,
+        2_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => {
+          const tab = findTab(handoff, secondTab.tabId);
+          return (
+            tab.status === "idle" &&
+            transcriptIncludesAgentText(tab.transcript, expectedReply)
+          );
+        },
+      );
+
+      const secondTranscript = findTab(withSecondReply, secondTab.tabId).transcript;
+      expect(transcriptIncludesAgentText(secondTranscript, expectedReply)).toBe(true);
+
+      await client.setWorkbenchSessionUnread(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: secondTab.tabId,
+        unread: false,
+      });
+      await client.markWorkbenchUnread(workspaceId, { handoffId: created.handoffId });
+
+      const unreadSnapshot = findHandoff(await client.getWorkbench(workspaceId), created.handoffId);
+      expect(unreadSnapshot.tabs.some((tab) => tab.unread)).toBe(true);
+
+      await client.closeWorkbenchSession(workspaceId, {
+        handoffId: created.handoffId,
+        tabId: secondTab.tabId,
+      });
+
+      const closedSnapshot = await poll(
+        "secondary session closed",
+        30_000,
+        1_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => !handoff.tabs.some((tab) => tab.id === secondTab.tabId),
+      );
+      expect(closedSnapshot.tabs).toHaveLength(1);
+
+      await client.revertWorkbenchFile(workspaceId, {
+        handoffId: created.handoffId,
+        path: expectedFile,
+      });
+
+      const revertedSnapshot = await poll(
+        "file revert reflected in workbench",
+        30_000,
+        1_000,
+        async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+        (handoff) => !handoff.fileChanges.some((file) => file.path === expectedFile),
+      );
+
+      expect(revertedSnapshot.fileChanges.some((file) => file.path === expectedFile)).toBe(false);
+      expect(revertedSnapshot.title).toBe(`Workbench E2E ${runId} Renamed`);
+      expect(findTab(revertedSnapshot, primaryTab.id).sessionName).toBe("Primary Session");
+    },
+  );
+});

factory/packages/client/test/e2e/workbench-load-e2e.test.ts

@@ -0,0 +1,331 @@
+import { describe, expect, it } from "vitest";
+import type {
+  HandoffWorkbenchSnapshot,
+  WorkbenchAgentTab,
+  WorkbenchHandoff,
+  WorkbenchModelId,
+  WorkbenchTranscriptEvent,
+} from "@openhandoff/shared";
+import { createBackendClient } from "../../src/backend-client.js";
+
+const RUN_WORKBENCH_LOAD_E2E = process.env.HF_ENABLE_DAEMON_WORKBENCH_LOAD_E2E === "1";
+
+function requiredEnv(name: string): string {
+  const value = process.env[name]?.trim();
+  if (!value) {
+    throw new Error(`Missing required env var: ${name}`);
+  }
+  return value;
+}
+
+function workbenchModelEnv(name: string, fallback: WorkbenchModelId): WorkbenchModelId {
+  const value = process.env[name]?.trim();
+  switch (value) {
+    case "claude-sonnet-4":
+    case "claude-opus-4":
+    case "gpt-4o":
+    case "o3":
+      return value;
+    default:
+      return fallback;
+  }
+}
+
+function intEnv(name: string, fallback: number): number {
+  const raw = process.env[name]?.trim();
+  if (!raw) {
+    return fallback;
+  }
+  const value = Number.parseInt(raw, 10);
+  return Number.isFinite(value) && value > 0 ? value : fallback;
+}
+
+async function sleep(ms: number): Promise<void> {
+  await new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+async function poll<T>(
+  label: string,
+  timeoutMs: number,
+  intervalMs: number,
+  fn: () => Promise<T>,
+  isDone: (value: T) => boolean,
+): Promise<T> {
+  const startedAt = Date.now();
+  let lastValue: T;
+
+  for (;;) {
+    lastValue = await fn();
+    if (isDone(lastValue)) {
+      return lastValue;
+    }
+    if (Date.now() - startedAt > timeoutMs) {
+      throw new Error(`timed out waiting for ${label}`);
+    }
+    await sleep(intervalMs);
+  }
+}
+
+function findHandoff(snapshot: HandoffWorkbenchSnapshot, handoffId: string): WorkbenchHandoff {
+  const handoff = snapshot.handoffs.find((candidate) => candidate.id === handoffId);
+  if (!handoff) {
+    throw new Error(`handoff ${handoffId} missing from snapshot`);
+  }
+  return handoff;
+}
+
+function findTab(handoff: WorkbenchHandoff, tabId: string): WorkbenchAgentTab {
+  const tab = handoff.tabs.find((candidate) => candidate.id === tabId);
+  if (!tab) {
+    throw new Error(`tab ${tabId} missing from handoff ${handoff.id}`);
+  }
+  return tab;
+}
+
+function extractEventText(event: WorkbenchTranscriptEvent): string {
+  const payload = event.payload;
+  if (!payload || typeof payload !== "object") {
+    return String(payload ?? "");
+  }
+
+  const envelope = payload as {
+    method?: unknown;
+    params?: unknown;
+    result?: unknown;
+  };
+
+  const params = envelope.params;
+  if (params && typeof params === "object") {
+    const update = (params as { update?: unknown }).update;
+    if (update && typeof update === "object") {
+      const content = (update as { content?: unknown }).content;
+      if (content && typeof content === "object") {
+        const chunkText = (content as { text?: unknown }).text;
+        if (typeof chunkText === "string") {
+          return chunkText;
+        }
+      }
+    }
+
+    const text = (params as { text?: unknown }).text;
+    if (typeof text === "string" && text.trim()) {
+      return text.trim();
+    }
+
+    const prompt = (params as { prompt?: Array<{ text?: unknown }> }).prompt;
+    if (Array.isArray(prompt)) {
+      return prompt
+        .map((item) => (typeof item?.text === "string" ? item.text.trim() : ""))
+        .filter(Boolean)
+        .join("\n");
+    }
+  }
+
+  const result = envelope.result;
+  if (result && typeof result === "object") {
+    const text = (result as { text?: unknown }).text;
+    if (typeof text === "string" && text.trim()) {
+      return text.trim();
+    }
+  }
+
+  return typeof envelope.method === "string" ? envelope.method : JSON.stringify(payload);
+}
+
+function transcriptIncludesAgentText(transcript: WorkbenchTranscriptEvent[], expectedText: string): boolean {
+  return transcript
+    .filter((event) => event.sender === "agent")
+    .map((event) => extractEventText(event))
+    .join("")
+    .includes(expectedText);
+}
+
+function average(values: number[]): number {
+  return values.reduce((sum, value) => sum + value, 0) / Math.max(values.length, 1);
+}
+
+async function measureWorkbenchSnapshot(
+  client: ReturnType<typeof createBackendClient>,
+  workspaceId: string,
+  iterations: number,
+): Promise<{
+  avgMs: number;
+  maxMs: number;
+  payloadBytes: number;
+  handoffCount: number;
+  tabCount: number;
+  transcriptEventCount: number;
+}> {
+  const durations: number[] = [];
+  let snapshot: HandoffWorkbenchSnapshot | null = null;
+
+  for (let index = 0; index < iterations; index += 1) {
+    const startedAt = performance.now();
+    snapshot = await client.getWorkbench(workspaceId);
+    durations.push(performance.now() - startedAt);
+  }
+
+  const finalSnapshot = snapshot ?? {
+    workspaceId,
+    repos: [],
+    projects: [],
+    handoffs: [],
+  };
+  const payloadBytes = Buffer.byteLength(JSON.stringify(finalSnapshot), "utf8");
+  const tabCount = finalSnapshot.handoffs.reduce((sum, handoff) => sum + handoff.tabs.length, 0);
+  const transcriptEventCount = finalSnapshot.handoffs.reduce(
+    (sum, handoff) =>
+      sum + handoff.tabs.reduce((tabSum, tab) => tabSum + tab.transcript.length, 0),
+    0,
+  );
+
+  return {
+    avgMs: Math.round(average(durations)),
+    maxMs: Math.round(Math.max(...durations, 0)),
+    payloadBytes,
+    handoffCount: finalSnapshot.handoffs.length,
+    tabCount,
+    transcriptEventCount,
+  };
+}
+
+describe("e2e(client): workbench load", () => {
+  it.skipIf(!RUN_WORKBENCH_LOAD_E2E)(
+    "runs a simple sequential load profile against the real backend",
+    { timeout: 30 * 60_000 },
+    async () => {
+      const endpoint = process.env.HF_E2E_BACKEND_ENDPOINT?.trim() || "http://127.0.0.1:7741/api/rivet";
+      const workspaceId = process.env.HF_E2E_WORKSPACE?.trim() || "default";
+      const repoRemote = requiredEnv("HF_E2E_GITHUB_REPO");
+      const model = workbenchModelEnv("HF_E2E_MODEL", "gpt-4o");
+      const handoffCount = intEnv("HF_LOAD_HANDOFF_COUNT", 3);
+      const extraSessionCount = intEnv("HF_LOAD_EXTRA_SESSION_COUNT", 2);
+      const pollIntervalMs = intEnv("HF_LOAD_POLL_INTERVAL_MS", 2_000);
+
+      const client = createBackendClient({
+        endpoint,
+        defaultWorkspaceId: workspaceId,
+      });
+
+      const repo = await client.addRepo(workspaceId, repoRemote);
+      const createHandoffLatencies: number[] = [];
+      const provisionLatencies: number[] = [];
+      const createSessionLatencies: number[] = [];
+      const messageRoundTripLatencies: number[] = [];
+      const snapshotSeries: Array<{
+        handoffCount: number;
+        avgMs: number;
+        maxMs: number;
+        payloadBytes: number;
+        tabCount: number;
+        transcriptEventCount: number;
+      }> = [];
+
+      snapshotSeries.push(await measureWorkbenchSnapshot(client, workspaceId, 2));
+
+      for (let handoffIndex = 0; handoffIndex < handoffCount; handoffIndex += 1) {
+        const runId = `load-${handoffIndex}-${Date.now().toString(36)}`;
+        const initialReply = `LOAD_INIT_${runId}`;
+
+        const createStartedAt = performance.now();
+        const created = await client.createWorkbenchHandoff(workspaceId, {
+          repoId: repo.repoId,
+          title: `Workbench Load ${runId}`,
+          branch: `load/${runId}`,
+          model,
+          task: `Reply with exactly: ${initialReply}`,
+        });
+        createHandoffLatencies.push(performance.now() - createStartedAt);
+
+        const provisionStartedAt = performance.now();
+        const provisioned = await poll(
+          `handoff ${runId} provisioning`,
+          12 * 60_000,
+          pollIntervalMs,
+          async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+          (handoff) => {
+            const tab = handoff.tabs[0];
+            return Boolean(
+              tab &&
+                handoff.status === "idle" &&
+                tab.status === "idle" &&
+                transcriptIncludesAgentText(tab.transcript, initialReply),
+            );
+          },
+        );
+        provisionLatencies.push(performance.now() - provisionStartedAt);
+
+        expect(provisioned.tabs.length).toBeGreaterThan(0);
+        const primaryTab = provisioned.tabs[0]!;
+        expect(transcriptIncludesAgentText(primaryTab.transcript, initialReply)).toBe(true);
+
+        for (let sessionIndex = 0; sessionIndex < extraSessionCount; sessionIndex += 1) {
+          const expectedReply = `LOAD_REPLY_${runId}_${sessionIndex}`;
+          const createSessionStartedAt = performance.now();
+          const createdSession = await client.createWorkbenchSession(workspaceId, {
+            handoffId: created.handoffId,
+            model,
+          });
+          createSessionLatencies.push(performance.now() - createSessionStartedAt);
+
+          await client.sendWorkbenchMessage(workspaceId, {
+            handoffId: created.handoffId,
+            tabId: createdSession.tabId,
+            text: `Run pwd in the repo, then reply with exactly: ${expectedReply}`,
+            attachments: [],
+          });
+
+          const messageStartedAt = performance.now();
+          const withReply = await poll(
+            `handoff ${runId} session ${sessionIndex} reply`,
+            10 * 60_000,
+            pollIntervalMs,
+            async () => findHandoff(await client.getWorkbench(workspaceId), created.handoffId),
+            (handoff) => {
+              const tab = findTab(handoff, createdSession.tabId);
+              return tab.status === "idle" && transcriptIncludesAgentText(tab.transcript, expectedReply);
+            },
+          );
+          messageRoundTripLatencies.push(performance.now() - messageStartedAt);
+
+          expect(transcriptIncludesAgentText(findTab(withReply, createdSession.tabId).transcript, expectedReply)).toBe(true);
+        }
+
+        const snapshotMetrics = await measureWorkbenchSnapshot(client, workspaceId, 3);
+        snapshotSeries.push(snapshotMetrics);
+        console.info(
+          "[workbench-load-snapshot]",
+          JSON.stringify({
+            handoffIndex: handoffIndex + 1,
+            ...snapshotMetrics,
+          }),
+        );
+      }
+
+      const firstSnapshot = snapshotSeries[0]!;
+      const lastSnapshot = snapshotSeries[snapshotSeries.length - 1]!;
+      const summary = {
+        handoffCount,
+        extraSessionCount,
+        createHandoffAvgMs: Math.round(average(createHandoffLatencies)),
+        provisionAvgMs: Math.round(average(provisionLatencies)),
+        createSessionAvgMs: Math.round(average(createSessionLatencies)),
+        messageRoundTripAvgMs: Math.round(average(messageRoundTripLatencies)),
+        snapshotReadBaselineAvgMs: firstSnapshot.avgMs,
+        snapshotReadFinalAvgMs: lastSnapshot.avgMs,
+        snapshotReadFinalMaxMs: lastSnapshot.maxMs,
+        snapshotPayloadBaselineBytes: firstSnapshot.payloadBytes,
+        snapshotPayloadFinalBytes: lastSnapshot.payloadBytes,
+        snapshotTabFinalCount: lastSnapshot.tabCount,
+        snapshotTranscriptFinalCount: lastSnapshot.transcriptEventCount,
+      };
+
+      console.info("[workbench-load-summary]", JSON.stringify(summary));
+
+      expect(createHandoffLatencies.length).toBe(handoffCount);
+      expect(provisionLatencies.length).toBe(handoffCount);
+      expect(createSessionLatencies.length).toBe(handoffCount * extraSessionCount);
+      expect(messageRoundTripLatencies.length).toBe(handoffCount * extraSessionCount);
+    },
+  );
+});

factory/packages/client/test/keys.test.ts

@@ -0,0 +1,31 @@
+import { describe, expect, it } from "vitest";
+import {
+  handoffKey,
+  handoffStatusSyncKey,
+  historyKey,
+  projectBranchSyncKey,
+  projectKey,
+  projectPrSyncKey,
+  sandboxInstanceKey,
+  workspaceKey
+} from "../src/keys.js";
+
+describe("actor keys", () => {
+  it("prefixes every key with workspace namespace", () => {
+    const keys = [
+      workspaceKey("default"),
+      projectKey("default", "repo"),
+      handoffKey("default", "repo", "handoff"),
+      sandboxInstanceKey("default", "daytona", "sbx"),
+      historyKey("default", "repo"),
+      projectPrSyncKey("default", "repo"),
+      projectBranchSyncKey("default", "repo"),
+      handoffStatusSyncKey("default", "repo", "handoff", "sandbox-1", "session-1")
+    ];
+
+    for (const key of keys) {
+      expect(key[0]).toBe("ws");
+      expect(key[1]).toBe("default");
+    }
+  });
+});

factory/packages/client/test/view-model.test.ts

@@ -0,0 +1,92 @@
+import { describe, expect, it } from "vitest";
+import type { HandoffRecord } from "@openhandoff/shared";
+import {
+  filterHandoffs,
+  formatRelativeAge,
+  fuzzyMatch,
+  summarizeHandoffs
+} from "../src/view-model.js";
+
+const sample: HandoffRecord = {
+  workspaceId: "default",
+  repoId: "repo-a",
+  repoRemote: "https://example.com/repo-a.git",
+  handoffId: "handoff-1",
+  branchName: "feature/test",
+  title: "Test Title",
+  task: "Do test",
+  providerId: "daytona",
+  status: "running",
+  statusMessage: null,
+  activeSandboxId: "sandbox-1",
+  activeSessionId: "session-1",
+  sandboxes: [
+    {
+      sandboxId: "sandbox-1",
+      providerId: "daytona",
+      sandboxActorId: null,
+      switchTarget: "daytona://sandbox-1",
+      cwd: null,
+      createdAt: 1,
+      updatedAt: 1
+    }
+  ],
+  agentType: null,
+  prSubmitted: false,
+  diffStat: null,
+  prUrl: null,
+  prAuthor: null,
+  ciStatus: null,
+  reviewStatus: null,
+  reviewer: null,
+  conflictsWithMain: null,
+  hasUnpushed: null,
+  parentBranch: null,
+  createdAt: 1,
+  updatedAt: 1
+};
+
+describe("search helpers", () => {
+  it("supports ordered fuzzy matching", () => {
+    expect(fuzzyMatch("feature/test-branch", "ftb")).toBe(true);
+    expect(fuzzyMatch("feature/test-branch", "fbt")).toBe(false);
+  });
+
+  it("filters rows across branch and title", () => {
+    const rows: HandoffRecord[] = [
+      sample,
+      {
+        ...sample,
+        handoffId: "handoff-2",
+        branchName: "docs/update-intro",
+        title: "Docs Intro Refresh",
+        status: "idle"
+      }
+    ];
+    expect(filterHandoffs(rows, "doc")).toHaveLength(1);
+    expect(filterHandoffs(rows, "h2")).toHaveLength(1);
+    expect(filterHandoffs(rows, "test")).toHaveLength(2);
+  });
+});
+
+describe("summary helpers", () => {
+  it("formats relative age", () => {
+    expect(formatRelativeAge(9_000, 10_000)).toBe("1s");
+    expect(formatRelativeAge(0, 120_000)).toBe("2m");
+  });
+
+  it("summarizes by status and provider", () => {
+    const rows: HandoffRecord[] = [
+      sample,
+      { ...sample, handoffId: "handoff-2", status: "idle", providerId: "daytona" },
+      { ...sample, handoffId: "handoff-3", status: "error", providerId: "daytona" }
+    ];
+
+    const summary = summarizeHandoffs(rows);
+    expect(summary.total).toBe(3);
+    expect(summary.byStatus.running).toBe(1);
+    expect(summary.byStatus.idle).toBe(1);
+    expect(summary.byStatus.error).toBe(1);
+    expect(summary.byProvider.daytona).toBe(3);
+  });
+});