pi-ui-extend 1.0.32 → 1.0.36

package/external/pi-tools-suite/src/async-subagents/async-subagents.sample.jsonc

@@ -5,7 +5,7 @@
   // Global async-subagents routing.
   // Rule of thumb:
   // - GLM: cheap/fast discovery, grep-style scanning, docs/test inventory.
-  // - Gemini: frontend/UI/UX, visual/product polish, and multimodal-heavy analysis.
+  // - GLM-5.3 Flash: frontend/UI/UX, visual/product polish, and multimodal-heavy analysis.
   // - GPT/Claude: review/audit, deep reasoning, tests, and implementation/refactoring.
   // Override exact model IDs here if your pi provider uses different names.
   "defaultType": "quick",
@@ -34,7 +34,11 @@
   // screenshot inspection should use lookup; this only keeps guidance honest.
   // Matching is case-insensitive and supports `*` wildcards.
   "vision": {
-    "blindModelPatterns": ["zai/glm*", "glm*", "*/glm*"]
+    "blindModelPatterns": [
+      "zai/glm-4.5*", "glm-4.5*", "*/glm-4.5*",
+      "zai/glm-5-turbo*", "glm-5-turbo*", "*/glm-5-turbo*",
+      "zai/glm-5.3", "glm-5.3", "*/glm-5.3"
+    ]
   },
 
   // Optional per-type prompt customization (leave commented until needed):
@@ -109,14 +113,14 @@
 
   "presets": {
     "cheap": {
-      "description": "Use cheap GLM/Gemini Flash models for text/code roles.",
+      "description": "Use GLM models by role, including GLM-5.3 Flash for multimodal work.",
       "types": {
         "quick": { "model": "zai/glm-4.5-air", "thinking": "off" },
         "scan": { "model": "zai/glm-4.5-air", "thinking": "off" },
         "research": { "model": "zai/glm-5-turbo", "thinking": "low" },
         "docs": { "model": "zai/glm-4.5-air", "thinking": "low" },
-        "frontend": { "model": "antigravity/gemini-3-flash-preview", "fallbackModels": ["zai/glm-5.3"], "thinking": "medium" },
-        "browser-qa": { "model": "openai-codex/gpt-5.6-luna", "fallbackModels": ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"], "thinking": "low" },
+        "frontend": { "model": "zai/glm-5.3-flash", "thinking": "medium" },
+        "browser-qa": { "model": "zai/glm-5.3-flash", "fallbackModels": ["openai-codex/gpt-5.6-luna"], "thinking": "low" },
         "tests": { "model": "zai/glm-5-turbo", "thinking": "medium" },
         "review": { "model": "zai/glm-5.3", "thinking": "high" },
         "implement": { "model": "zai/glm-5.3", "thinking": "high" },
@@ -132,7 +136,7 @@
         "research": { "model": "openai-codex/gpt-5.6-terra", "fallbackModels": ["zai/glm-5-turbo"], "thinking": "low" },
         "docs": { "model": "openai-codex/gpt-5.6-luna", "fallbackModels": ["zai/glm-4.5-air"], "thinking": "low" },
         "frontend": { "model": "openai-codex/gpt-5.6-terra", "fallbackModels": ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"], "thinking": "medium" },
-        "browser-qa": { "model": "openai-codex/gpt-5.6-luna", "fallbackModels": ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"], "thinking": "low" },
+        "browser-qa": { "model": "zai/glm-5.3-flash", "fallbackModels": ["openai-codex/gpt-5.6-luna"], "thinking": "low" },
         "tests": { "model": "openai-codex/gpt-5.6-terra", "fallbackModels": ["zai/glm-5-turbo"], "thinking": "medium" },
         "review": { "model": "openai-codex/gpt-5.6-sol", "fallbackModels": ["zai/glm-5.3"], "thinking": "high" },
         "implement": { "model": "openai-codex/gpt-5.6-sol", "fallbackModels": ["zai/glm-5.3"], "thinking": "high" },
@@ -148,7 +152,7 @@
         "research": { "model": "antigravity/gemini-3.1-pro-preview", "fallbackModels": ["openai-codex/gpt-5.6-luna", "zai/glm-5-turbo"], "thinking": "medium" },
         "docs": { "model": "antigravity/gemini-2.5-flash", "fallbackModels": ["openai-codex/gpt-5.6-luna", "zai/glm-4.5-air"], "thinking": "medium" },
         "frontend": { "model": "antigravity/gemini-3.1-pro-preview-customtools", "fallbackModels": ["openai-codex/gpt-5.6-luna", "zai/glm-5.3"], "thinking": "low" },
-        "browser-qa": { "model": "openai-codex/gpt-5.6-luna", "fallbackModels": ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"], "thinking": "low" },
+        "browser-qa": { "model": "zai/glm-5.3-flash", "fallbackModels": ["openai-codex/gpt-5.6-luna"], "thinking": "low" },
         "tests": { "model": "antigravity/antigravity-claude-sonnet-4-6", "fallbackModels": ["openai-codex/gpt-5.6-luna", "zai/glm-5-turbo"], "thinking": "high" },
         "review": { "model": "antigravity/antigravity-claude-sonnet-4-6", "fallbackModels": ["openai-codex/gpt-5.6-sol", "zai/glm-5.3"], "thinking": "high" },
         "implement": { "model": "openai-codex/gpt-5.6-sol", "fallbackModels": ["zai/glm-5.3"], "thinking": "high" },
@@ -203,8 +207,8 @@
 
     "browser-qa": {
       "description": "Use for browser-based visual QA: reproduce UI bugs and verify fixes with deterministic assertions, screenshots, video, and traces.",
-      "model": "openai-codex/gpt-5.6-luna",
-      "fallbackModels": ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"],
+      "model": "zai/glm-5.3-flash",
+      "fallbackModels": ["openai-codex/gpt-5.6-luna"],
       "thinking": "low",
       "timeoutMs": 120000,
       "tools": ["read", "grep", "bash"]

package/external/pi-tools-suite/src/async-subagents/core/config.ts

@@ -215,8 +215,8 @@ const BUILTIN_CONFIG: SubagentConfig = {
 		},
 		"browser-qa": {
 			description: "Use for browser-based visual QA: reproduce UI bugs and verify fixes with deterministic assertions, screenshots, video, and traces.",
-			model: "openai-codex/gpt-5.6-luna",
-			fallbackModels: ["antigravity/gemini-3-flash-preview", "zai/glm-5.3"],
+			model: "zai/glm-5.3-flash",
+			fallbackModels: ["openai-codex/gpt-5.6-luna"],
 			thinking: "low",
 			timeoutMs: 120_000,
 			tools: ["read", "grep", "bash"],

package/external/pi-tools-suite/src/async-subagents/core/spawn.ts

@@ -88,15 +88,18 @@ export function spawnAgent(
 	// detached subprocesses do not depend on persisted local pi settings.
 	const persistSessions = shouldPersistSubagentSessions();
 	const sessionDir = persistSessions ? getAgentSessionDir(agentDir) : undefined;
+	const forwardedExtraArgs = options.isolatedSkills?.length ? withoutSkillArgs(extraArgs) : extraArgs;
 	if (sessionDir) fs.mkdirSync(sessionDir, { recursive: true });
 	const piArgs: string[] = ["--mode", "rpc"];
 	if (sessionDir) piArgs.push("--session-dir", sessionDir);
 	else piArgs.push("--no-session");
 	piArgs.push("--no-extensions");
 	piArgs.push("--extension", getModelToolsExtensionPath());
-	// `--no-extensions` keeps sub-agents isolated, but the suite-owned provider
-	// is infrastructure: always restore it, regardless of the selected model.
-	piArgs.push("--extension", getAntigravityAuthExtensionPath());
+	// Preserve `--no-extensions` unless this invocation explicitly selects an
+	// Antigravity model. Environment/default models do not opt the provider in.
+	if (usesAntigravityModel(task.model, forwardedExtraArgs)) {
+		piArgs.push("--extension", getAntigravityAuthExtensionPath());
+	}
 	if (options.isolatedSkills && options.isolatedSkills.length > 0) {
 		piArgs.push("--no-skills");
 		for (const skillPath of options.isolatedSkills) piArgs.push("--skill", skillPath);
@@ -111,7 +114,7 @@ export function spawnAgent(
 	if (task.thinking) piArgs.push("--thinking", task.thinking);
 
 	// User-supplied extra args (e.g. --thinking high)
-	piArgs.push(...(options.isolatedSkills?.length ? withoutSkillArgs(extraArgs) : extraArgs));
+	piArgs.push(...forwardedExtraArgs);
 	// Keep recursive/interactive parent-only tools disabled even if explicit
 	// sub-agent extraArgs load additional extensions or override --tools.
 	piArgs.push("--extension", getSubagentToolGuardExtensionPath());
@@ -470,6 +473,22 @@ function withoutSkillArgs(args: string[]): string[] {
 	return filtered;
 }
 
+function usesAntigravityModel(taskModel: string | undefined, extraArgs: string[]): boolean {
+	let model = taskModel?.trim() || undefined;
+	for (let index = 0; index < extraArgs.length; index += 1) {
+		const arg = extraArgs[index];
+		if (arg === "--model" || arg === "-m") {
+			model = extraArgs[index + 1]?.trim() || undefined;
+			index += 1;
+			continue;
+		}
+		if (arg.startsWith("--model=")) {
+			model = arg.slice("--model=".length).trim() || undefined;
+		}
+	}
+	return Boolean(model?.startsWith("antigravity/") && model.length > "antigravity/".length);
+}
+
 function getModelToolsExtensionPath(): string {
 	return path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..", "..", "model-tools", "index.ts");
 }

package/external/pi-tools-suite/src/async-subagents/private-skills/browser-qa/SKILL.md

@@ -13,6 +13,15 @@ CLI, create shared/default browser sessions, or generate executable browser code
 Never read, print, grep, copy, or edit credential values from
 `.pi/qa_auth.jsonc` yourself.
 
+Treat the launch brief as a user-visible acceptance contract, not an execution
+plan. It should identify the actual target URL/app when known, the user flow,
+the expected observable result, and required evidence. Missing details are
+preflight unknowns, not permission to invent a different target. Never create,
+serve, or switch to a mock/synthetic page, test fixture, or built-in harness
+unless the user explicitly requested that exact target. Source inspection and
+repository tests may support discovery, but they never substitute for testing
+the requested target in the browser.
+
 ## Workflow
 
 Treat target discovery as a 30-second preflight and invoke the runner within 45
@@ -27,9 +36,11 @@ source reading, server polling, capability probing, or retries.
    The launcher creates its private `flows/` directory and the runner rejects
    flows or evidence destinations outside this owning sub-agent directory. Do
    not override `PI_SUBAGENT_AGENT_DIR` or copy evidence to shared project paths.
-3. Discover the requested target, expected behavior, and the smallest scenario
-   that can prove it. If the target cannot be reached or started, report the
-   concrete blocker instead of substituting static checks for browser QA.
+3. Discover the requested actual target, expected behavior, and the smallest
+   scenario that can prove it. Treat parent-supplied repository details as hints
+   unless the user explicitly requested that exact harness. If the target cannot
+   be reached or started, report the concrete blocker instead of switching to a
+   mock target or substituting static checks for browser QA.
 4. If the requested behavior requires authentication, run
    `node <runner> profiles`. A missing auth config is valid and returns an empty
    list without creating `.pi/qa_auth.jsonc`. Otherwise skip profile discovery
@@ -38,7 +49,8 @@ source reading, server polling, capability probing, or retries.
    the requested page requires login. If form authentication is required but
    there is no usable profile, follow **Form-auth scaffolding** below instead of
    asking the user to discover selectors.
-5. Inspect the target code and write a declarative JSONC flow under
+5. Inspect only enough target code to identify a supported launch path or stable
+   locators, then write a declarative JSONC flow under
    `$PI_SUBAGENT_AGENT_DIR/browser-qa/flows/`. Never put credentials or raw
    executable JavaScript in it. The `evaluate` action exposes only the safe
    operations documented below; it does not accept expressions or scripts.
@@ -71,6 +83,23 @@ source reading, server polling, capability probing, or retries.
   and `waitForTimeout` only for short input settling or an unavoidable
   animation/debounce. Set flow `timeoutMs` only as high as the target
   legitimately needs.
+- The runner waits after every visible interaction until the document is ready,
+  requests started by the interaction have finished, and common visible busy
+  markers (including `aria-busy`, progress bars, loading/spinner/skeleton test
+  ids and classes) disappear. It then keeps the stable state on video for 500
+  ms. A busy page that does not settle within `timeoutMs` fails instead of
+  continuing against a skeleton. For an app-specific loader not covered by
+  those conventions, add an explicit `waitFor`/`assertHidden` for that loader
+  and assert the loaded content before interacting with it.
+- Recorded pointer actions are annotated automatically: clicks and double-clicks
+  show a cursor and pulse. After a native `dragTo` completes, the runner visibly
+  replays the resolved source-to-target route for 450 ms with a large orange
+  cursor and progressively drawn high-contrast trail, then shows a green drop
+  marker. The
+  isolated annotation layer applies to the main page, same-origin frames,
+  declared popups, and form-auth submission; it is accessibility-hidden, ignores
+  pointer input, and clears before the runner's post-action stable interval
+  completes.
 - Place `authRejectedIf` immediately after navigation or any transition that
   may reveal expired authentication.
 - Never weaken an assertion merely to make a failing run pass. If the observed
@@ -114,6 +143,13 @@ The top-level `environment` may set `locale`, `timezoneId`, `colorScheme`, and
 motion accepts `reduce` or `no-preference`. The resolved environment is returned
 in the result alongside the viewport.
 
+Navigation and visible interaction actions automatically wait for page
+readiness and a 500 ms stable recording interval. This applies to the main
+page, same-origin frames and declared popups, and to the trusted form-auth
+sequence. An explicit `waitForTimeout` is not extended by another automatic
+delay. Click-like actions use a short bounded press duration so their automatic
+video pulse remains visible even when the click immediately navigates.
+
 Triggering interactions (`click`, `doubleClick`, `press`, `check`, `uncheck`,
 and `selectOption`) may declare race-free expectations that are armed before
 the interaction:

package/external/pi-tools-suite/src/async-subagents/private-skills/browser-qa/references/qa-design.md

@@ -41,6 +41,15 @@ Runner interactions inherit Playwright auto-waiting. Usually an action followed
 by an assertion is enough. Use `waitFor` only when the next operation depends on
 a distinct attached/detached/visible/hidden transition.
 
+After navigation and visible interactions, the runner also waits for DOM
+readiness, completion of requests started by that action, disappearance of
+common visible `aria-busy`/progress/loading/spinner/skeleton markers, and a
+500 ms stable interval. If those signals remain busy through the flow timeout,
+the run fails rather than interacting with a loading shell. This is a safe
+baseline, not an application-specific oracle: explicitly wait for a custom
+loader to become hidden and assert the loaded content when the application uses
+different readiness semantics.
+
 `waitForTimeout` is bounded to five seconds and should be exceptional—for a
 known animation, debounce, or externally scheduled transition with no
 observable intermediate state. Sleeping longer hides races instead of proving
@@ -155,6 +164,16 @@ Use evidence by purpose:
 - **Video:** chronological confirmation of the complete user flow.
 - **Trace:** action/DOM timing diagnosis for a failed or flaky interaction.
 
+Videos automatically show a transient cursor and yellow pulse for clicks and
+double-clicks. After a native `dragTo` gesture completes, its resolved
+source-to-target route is replayed over 450 ms with a large orange cursor and
+progressively drawn high-contrast trail, followed by a green drop marker. These
+annotations are
+runner-owned, pointer-transparent, and accessibility-hidden; they cover the main
+page, same-origin frames, declared popups, and form-auth submission. Their
+bounded animations finish within the normal post-action stable interval, so
+they explain the chronology without becoming screenshot or assertion oracles.
+
 Assertions determine pass/fail; evidence explains it. Preserve and link every
 artifact group returned on both passed and failed runs.