@opengeni/react 3.1.0 → 3.1.2

package/src/terminal.ts

@@ -0,0 +1,26 @@
+import {
+  registerSandboxTerminal,
+  type SandboxTerminalPeers,
+  type WorkbenchPeerLoader,
+} from "./lib/workbench-peers";
+
+export { SandboxTerminal } from "./components/sandbox-terminal";
+export type { SandboxTerminalProps, XtermTheme } from "./components/sandbox-terminal";
+export { registerSandboxTerminal } from "./lib/workbench-peers";
+
+async function loadTerminal(): Promise<SandboxTerminalPeers> {
+  const [{ Terminal }, { FitAddon }, { WebLinksAddon }] = await Promise.all([
+    import("@xterm/xterm"),
+    import("@xterm/addon-fit"),
+    import("@xterm/addon-web-links"),
+  ]);
+  return { Terminal, FitAddon, WebLinksAddon };
+}
+
+/** Opt in once on a terminal route; WebGL remains a separate optional import. */
+export function enableSandboxTerminal(options?: { webgl?: WorkbenchPeerLoader | undefined }): void {
+  const loadWebgl = options?.webgl;
+  registerSandboxTerminal(
+    loadWebgl ? async () => ({ ...(await loadTerminal()), loadWebgl }) : loadTerminal,
+  );
+}

package/src/testing/session-scenarios.ts

@@ -0,0 +1,404 @@
+// Deterministic session scenarios for visual harnesses (web dev gallery and native lab).
+// They use the same durable event shapes a live session produces, so every renderer
+// replays identical input through the official projection with no model calls.
+import type { SessionEvent } from "@opengeni/sdk";
+
+export type LabScenarioId =
+  | "answer"
+  | "working"
+  | "needs-you"
+  | "history"
+  | "failure"
+  | "long"
+  | "media"
+  | "previewing";
+
+export interface LabScenario {
+  id: LabScenarioId;
+  title: string;
+  events: SessionEvent[];
+  /** Whether the session is currently running a turn. */
+  running: boolean;
+}
+
+function recorder(sessionId: string) {
+  const events: SessionEvent[] = [];
+  let clock = Date.parse("2026-10-03T09:00:00Z");
+  const push = (type: string, payload: unknown, turnId: string | null, advanceMs = 2_000) => {
+    clock += advanceMs;
+    events.push({
+      id: `${sessionId}-evt-${events.length + 1}`,
+      workspaceId: "lab-workspace",
+      sessionId,
+      sequence: events.length + 1,
+      type,
+      payload,
+      turnId,
+      occurredAt: new Date(clock).toISOString(),
+    } as SessionEvent);
+  };
+  const tool = (
+    turnId: string,
+    id: string,
+    name: string,
+    args: unknown,
+    output?: unknown,
+    durationMs = 3_000,
+  ) => {
+    push("agent.toolCall.created", { id, name, arguments: args }, turnId, 800);
+    if (output !== undefined) push("agent.toolCall.output", { id, output }, turnId, durationMs);
+  };
+  return { events, push, tool };
+}
+
+function answer(): LabScenario {
+  const r = recorder("lab-answer");
+  const t = "turn-answer";
+  r.push(
+    "user.message",
+    { text: "What's the difference between a retry and a resume for a failed run?" },
+    t,
+  );
+  r.push("turn.started", {}, t, 400);
+  r.push(
+    "agent.reasoning.delta",
+    { text: "The user wants a crisp conceptual distinction, with an example." },
+    t,
+    1_500,
+  );
+  const text = [
+    "**Retry** starts the failed step again from its last durable checkpoint. **Resume** continues the whole run from where it paused, keeping everything already done.",
+    "",
+    "### When to use which",
+    "",
+    "- **Retry** when a single step failed for a transient reason (timeout, rate limit).",
+    "- **Resume** when the run was *waiting* — for an approval, an answer, or capacity.",
+    "",
+    "```bash",
+    "opengeni sessions retry 7f3c --step build",
+    "```",
+    "",
+    "Neither replays side effects that already succeeded.",
+  ].join("\n");
+  r.push("agent.message.delta", { text }, t, 2_500);
+  r.push("agent.message.completed", { text }, t, 300);
+  r.push("turn.completed", { output: text }, t, 200);
+  return { id: "answer", title: "Quick answer", events: r.events, running: false };
+}
+
+function working(): LabScenario {
+  const r = recorder("lab-working");
+  const t = "turn-working";
+  r.push(
+    "user.message",
+    {
+      text: "The checkout tests started failing after the currency change. Can you find out why and fix it?",
+    },
+    t,
+  );
+  r.push("turn.started", {}, t, 400);
+  r.push(
+    "agent.reasoning.delta",
+    { text: "Start by reproducing the failure, then narrow to the currency formatting path." },
+    t,
+    1_800,
+  );
+  r.tool(
+    t,
+    "c1",
+    "exec_command",
+    { cmd: "bun test checkout" },
+    "3 failed, 41 passed\n✗ formats totals in NOK\n✗ rounds VAT per line\n✗ shows discount",
+    6_000,
+  );
+  r.tool(
+    t,
+    "c2",
+    "search_files",
+    { query: "formatCurrency(" },
+    "src/money/format.ts:12\nsrc/checkout/summary.tsx:48\nsrc/checkout/line.tsx:31",
+  );
+  r.push(
+    "agent.message.delta",
+    {
+      text: "Found it: the currency change switched to minor units, but the summary still divides by 100.",
+      phase: "commentary",
+    },
+    t,
+    1_200,
+  );
+  r.push(
+    "agent.message.completed",
+    {
+      text: "Found it: the currency change switched to minor units, but the summary still divides by 100.",
+      phase: "commentary",
+    },
+    t,
+    200,
+  );
+  r.tool(t, "c3", "apply_patch", { path: "src/checkout/summary.tsx" }, "Updated 1 file (+3 −5)");
+  r.tool(t, "c4", "exec_command", { cmd: "bun test checkout" });
+  return { id: "working", title: "Working", events: r.events, running: true };
+}
+
+function needsYou(): LabScenario {
+  const r = recorder("lab-needs-you");
+  const t = "turn-needs-you";
+  r.push("user.message", { text: "Ship the fix to the preview environment and tell the team." }, t);
+  r.push("turn.started", {}, t, 400);
+  r.tool(t, "n1", "exec_command", { cmd: "bun run build" }, "Build completed in 41s", 8_000);
+  r.tool(t, "n2", "list_channels", { query: "release" }, "#releases, #release-ops");
+  r.push(
+    "agent.toolCall.created",
+    {
+      id: "n3",
+      name: "deploy_preview",
+      arguments: { environment: "preview", ref: "fix/checkout-minor-units" },
+    },
+    t,
+    900,
+  );
+  r.push(
+    "session.requiresAction",
+    {
+      approvals: [
+        {
+          id: "n3",
+          name: "deploy_preview",
+          arguments: { environment: "preview", ref: "fix/checkout-minor-units" },
+        },
+      ],
+    },
+    t,
+    300,
+  );
+  r.push(
+    "session.humanInput.requested",
+    {
+      request: {
+        id: "hi-1",
+        allowSkip: true,
+        questions: [
+          {
+            id: "channel",
+            kind: "single_select",
+            prompt: "Where should I announce the preview?",
+            label: "Announcement",
+            required: true,
+            options: [
+              { id: "releases", label: "#releases", description: "Team-wide release notes" },
+              { id: "release-ops", label: "#release-ops", description: "Operators only" },
+            ],
+          },
+        ],
+      },
+    },
+    t,
+    300,
+  );
+  return { id: "needs-you", title: "Needs you", events: r.events, running: true };
+}
+
+function history(): LabScenario {
+  const r = recorder("lab-history");
+  const a = "turn-h1";
+  r.push("user.message", { text: "Summarise yesterday's support tickets." }, a);
+  r.push("turn.started", {}, a, 400);
+  r.tool(a, "h1", "search_tickets", { since: "yesterday" }, "38 tickets", 4_000);
+  r.tool(a, "h2", "read_ticket", { id: "T-1182" }, "Customer cannot export invoices to PDF");
+  r.tool(a, "h3", "read_ticket", { id: "T-1190" }, "Login loop on Android 15");
+  const summary =
+    "38 tickets yesterday. The two themes:\n\n1. **PDF export** fails for invoices with attachments (11 tickets).\n2. **Android login loop** after the last release (7 tickets).\n\nEverything else was routine.";
+  r.push("agent.message.delta", { text: summary }, a, 2_000);
+  r.push("agent.message.completed", { text: summary }, a, 200);
+  r.push("turn.completed", { output: summary }, a, 200);
+  const b = "turn-h2";
+  r.push(
+    "user.message",
+    { text: "Open an issue for the Android one and assign it to mobile." },
+    b,
+    60_000,
+  );
+  r.push("turn.started", {}, b, 400);
+  r.tool(
+    b,
+    "h4",
+    "create_issue",
+    { title: "Android login loop after release", team: "mobile" },
+    "Error: tracker token expired",
+    2_000,
+  );
+  r.push(
+    "turn.failed",
+    { error: "The issue tracker connection needs to be reauthorized." },
+    b,
+    400,
+  );
+  const c = "turn-h3";
+  r.push("user.message", { text: "Reconnected. Try again." }, c, 30_000);
+  r.push("turn.started", {}, c, 400);
+  r.tool(
+    c,
+    "h5",
+    "create_issue",
+    { title: "Android login loop after release", team: "mobile" },
+    "Created MOB-412",
+    2_000,
+  );
+  const done = "Created **MOB-412** and assigned it to the mobile team.";
+  r.push("agent.message.delta", { text: done }, c, 1_000);
+  r.push("agent.message.completed", { text: done }, c, 200);
+  r.push("turn.completed", { output: done }, c, 200);
+  return { id: "history", title: "History", events: r.events, running: false };
+}
+
+function failure(): LabScenario {
+  const r = recorder("lab-failure");
+  const t = "turn-failure";
+  r.push("user.message", { text: "Deploy the latest build to staging." }, t);
+  r.push("turn.started", {}, t, 400);
+  r.tool(t, "f1", "exec_command", { cmd: "bun run build" }, "Build completed in 38s", 7_000);
+  r.tool(
+    t,
+    "f2",
+    "exec_command",
+    { cmd: "deploy --env staging" },
+    "error: staging cluster unreachable (connection refused after 3 attempts)",
+    4_000,
+  );
+  r.push(
+    "turn.failed",
+    { error: "The staging cluster refused the connection. Nothing was deployed." },
+    t,
+    400,
+  );
+  return { id: "failure", title: "Failure", events: r.events, running: false };
+}
+
+function longContent(): LabScenario {
+  const r = recorder("lab-long");
+  const t = "turn-long";
+  const question = [
+    "Here is the incident timeline from the on-call channel. Please turn it into a short postmortem with a summary, impact, root cause, a table of the timeline and the follow-ups.",
+    "",
+    "09:02 alerts fire for elevated 5xx on the checkout API; 09:05 on-call acknowledges; 09:11 rollback of release 412 starts; 09:19 error rate back to baseline; 09:40 root cause identified as a missing index after the orders migration; 10:15 index added and release 412 redeployed.",
+  ].join("\n");
+  r.push("user.message", { text: question }, t);
+  r.push("turn.started", {}, t, 400);
+  r.push(
+    "agent.reasoning.delta",
+    { text: "Structure it as a standard postmortem with a timeline table." },
+    t,
+    1_500,
+  );
+  const text = [
+    "## Postmortem: checkout 5xx after release 412",
+    "",
+    "**Summary.** For 17 minutes the checkout API returned elevated server errors after release 412. Rolling back restored service; the fix shipped the same morning.",
+    "",
+    "**Impact.** Roughly 6% of checkout attempts failed between 09:02 and 09:19. No orders were lost, but some customers had to retry.",
+    "",
+    "**Root cause.** The orders migration in release 412 dropped and recreated a table without its `customer_id` index, so a hot query fell back to a sequential scan and timed out under load.",
+    "",
+    "### Timeline",
+    "",
+    "| Time | Event |",
+    "| --- | --- |",
+    "| 09:02 | Alerts fire for elevated 5xx on checkout |",
+    "| 09:05 | On-call acknowledges |",
+    "| 09:11 | Rollback of release 412 starts |",
+    "| 09:19 | Error rate back to baseline |",
+    "| 09:40 | Missing index identified |",
+    "| 10:15 | Index added, release 412 redeployed |",
+    "",
+    "### Follow-ups",
+    "",
+    "1. Add a migration check that fails when an index disappears.",
+    "2. Alert on query latency, not only error rate.",
+    "3. Document the rollback runbook link in the alert itself.",
+    "",
+    "```sql",
+    "CREATE INDEX CONCURRENTLY orders_customer_id_idx ON orders (customer_id);",
+    "```",
+    "",
+    "> Rollback was the right first move: it bought time to find the cause without pressure.",
+  ].join("\n");
+  r.push("agent.message.delta", { text }, t, 3_000);
+  r.push("agent.message.completed", { text }, t, 300);
+  r.push("turn.completed", { output: text }, t, 200);
+  return { id: "long", title: "Long content", events: r.events, running: false };
+}
+
+const PREVIEW_FENCE_BODY = [
+  '<div class="card" style="padding:16px">',
+  '  <h2 style="margin:0 0 8px">Weekly signups</h2>',
+  '  <p style="margin:0 0 12px;color:#777">Last 6 weeks, all channels</p>',
+  '  <div style="display:flex;gap:8px;align-items:flex-end;height:140px">',
+  ...[42, 58, 51, 73, 88, 96].map(
+    (value) =>
+      `    <div style="flex:1;height:${value}%;border-radius:6px;background:linear-gradient(#5fb8a3,#2f7d6d)"></div>`,
+  ),
+  "  </div>",
+  "</div>",
+].join("\n");
+
+/** Wide table, an image and a finished interactive preview. */
+function media(): LabScenario {
+  const r = recorder("lab-media");
+  const t = "turn-media";
+  r.push("user.message", { text: "Compare the plans and show me the signup trend." }, t);
+  r.push("turn.started", {}, t, 400);
+  const text = [
+    "Signups are up again. Here is the trend as a live chart:",
+    "",
+    "```opengeni-html",
+    PREVIEW_FENCE_BODY,
+    "```",
+    "",
+    "![Signup funnel](https://picsum.photos/id/1056/1200/675)",
+    "",
+    "And the comparison across every plan:",
+    "",
+    "| Plan | Monthly price | Seats included | Storage | Support | Uptime SLA | Overage per seat |",
+    "| --- | --- | --- | --- | --- | --- | --- |",
+    "| Starter | $0 | 3 | 5 GB | Community forum | None | Not available |",
+    "| Team | $49 | 10 | 100 GB | Email, next business day | 99.5% | $6 |",
+    "| Business | $199 | 50 | 1 TB | Chat and email, 4 hours | 99.9% | $5 |",
+    "| Enterprise | Custom | Unlimited | Custom | Dedicated manager | 99.99% | Negotiated |",
+    "",
+    "Signups grew every week except week 3.",
+  ].join("\n");
+  r.push("agent.message.delta", { text }, t, 2_500);
+  r.push("agent.message.completed", { text }, t, 300);
+  r.push("turn.completed", { output: text }, t, 200);
+  return { id: "media", title: "Tables & media", events: r.events, running: false };
+}
+
+/** The assistant is still writing an interactive preview. */
+function previewing(): LabScenario {
+  const r = recorder("lab-previewing");
+  const t = "turn-previewing";
+  r.push("user.message", { text: "Make me a quick chart of weekly signups." }, t);
+  r.push("turn.started", {}, t, 400);
+  const partial = PREVIEW_FENCE_BODY.split("\n").slice(0, 5).join("\n");
+  r.push(
+    "agent.message.delta",
+    { text: ["Here is the chart:", "", "```opengeni-html", partial].join("\n") },
+    t,
+    2_000,
+  );
+  return { id: "previewing", title: "Writing a preview", events: r.events, running: true };
+}
+
+export function labScenarios(): LabScenario[] {
+  return [
+    answer(),
+    working(),
+    needsYou(),
+    history(),
+    failure(),
+    longContent(),
+    media(),
+    previewing(),
+  ];
+}