e2e 0.21.1 → 0.21.2

package/docs/snippets/setup-prompt-card.jsx

@@ -0,0 +1,22 @@
+/** Shows a faded prompt preview with an accessible expand and collapse button. */
+export const SetupPrompt = ({ children }) => {
+  const [expanded, setExpanded] = useState(false);
+
+  return (
+    <div className="setup-prompt" data-expanded={expanded}>
+      <div className="setup-prompt-header">
+        <span>Agent prompt</span>
+      </div>
+      <div id="setup-prompt-content">{children}</div>
+      <button
+        type="button"
+        className="setup-prompt-toggle"
+        aria-expanded={expanded}
+        aria-controls="setup-prompt-content"
+        onClick={() => setExpanded(!expanded)}
+      >
+        {expanded ? 'Show less' : 'Show more'}
+      </button>
+    </div>
+  );
+};

package/docs/snippets/show-more.jsx

@@ -0,0 +1,22 @@
+/** Shows the start of its content in a hairline box, faded out over a "Show more" button that expands it. */
+export const ShowMore = ({ id, children }) => {
+  const [expanded, setExpanded] = useState(false);
+
+  return (
+    <div className="show-more" data-expanded={expanded}>
+      {/* Tabbing into the folded part opens it, so focus never lands on clipped text. */}
+      <div id={id} className="show-more-content" onFocus={() => setExpanded(true)}>
+        {children}
+      </div>
+      <button
+        type="button"
+        className="show-more-toggle"
+        aria-expanded={expanded}
+        aria-controls={id}
+        onClick={() => setExpanded(!expanded)}
+      >
+        {expanded ? 'Show less' : 'Show more'}
+      </button>
+    </div>
+  );
+};

package/docs/snippets/video-card.jsx

@@ -0,0 +1,34 @@
+/** Links to a YouTube video in a new tab: thumbnail on the left, title and subtitle on the right. */
+export const VideoCard = ({ id, title, subtitle }) => (
+  <a
+    className="video-card"
+    href={`https://www.youtube.com/watch?v=${id}`}
+    target="_blank"
+    rel="noopener noreferrer"
+  >
+    <span className="video-card-thumbnail">
+      <img src={`https://i.ytimg.com/vi/${id}/hqdefault.jpg`} alt="" loading="lazy" noZoom />
+      <svg className="video-card-play" viewBox="0 0 48 48" aria-hidden="true">
+        <circle cx="24" cy="24" r="24" fill="rgb(0 0 0 / 0.6)" />
+        <path d="M19 15.5v17l14-8.5z" fill="#fff" />
+      </svg>
+    </span>
+    <span className="video-card-text">
+      <span className="video-card-title">{title}</span>
+      <span className="video-card-subtitle">{subtitle}</span>
+      <span className="video-card-new-tab"> (opens in a new tab)</span>
+    </span>
+    <svg
+      className="video-card-arrow"
+      viewBox="0 0 24 24"
+      fill="none"
+      stroke="currentColor"
+      strokeWidth="2"
+      strokeLinecap="round"
+      strokeLinejoin="round"
+      aria-hidden="true"
+    >
+      <path d="M7 17 17 7M7 7h10v10" />
+    </svg>
+  </a>
+);

package/docs/snippets/video.jsx

@@ -0,0 +1,25 @@
+/**
+ * A muted, looping clip in a hairline frame with its caption underneath. It
+ * plays on its own unless the reader prefers reduced motion; the controls
+ * are always there.
+ *
+ * The page writes the `<video>` element itself, followed by the caption:
+ * Mintlify resolves the `src` of a `<video>` in MDX to the deployed asset,
+ * but never a prop passed to a custom component, so `src="/images/x.mp4"`
+ * as a prop 404s wherever the docs are served under a subpath (`/docs`).
+ */
+export const Video = ({ children }) => {
+  const figure = useRef(null);
+
+  useEffect(() => {
+    if (window.matchMedia('(prefers-reduced-motion: no-preference)').matches) {
+      figure.current?.querySelector('video')?.play().catch(() => {});
+    }
+  }, []);
+
+  return (
+    <div ref={figure} className="video-figure">
+      {children}
+    </div>
+  );
+};

package/docs/subscriptions.mdx

@@ -71,6 +71,8 @@ import { copilot } from 'e2e/oauth/copilot';
 
 const agent = { model: copilot('claude-sonnet-5') };
 ```
+
+Which models your plan serves is up to GitHub; see [Pick a model](#pick-a-model).
   </Tab>
   <Tab title="OpenCode Console">
 <CodeGroup>
@@ -191,9 +193,14 @@ The id is not always the name the vendor's app shows.
 With `OPENCODE_API_KEY` set, `e2e models opencode-console` lists the ids that
 key serves, the same key `opencodeConsole()` uses in place of the login.
 
-- **Copilot** marks the models `copilot()` can't call. For `not enabled`,
-  turn the model on in your Copilot settings on GitHub, or ask the admin who
-  manages your seat. For `no chat or responses`, use a different model.
+- **Copilot** marks `not enabled` (turn the model on in your Copilot
+  settings on GitHub, or ask the admin who manages your seat) and `no chat
+  or responses` (use a different model). It can't see what your plan
+  serves: on Copilot Free, GitHub serves models through
+  [auto model selection only](https://docs.github.com/en/copilot/get-started/plans),
+  so a model called by id is usually refused. The run then stops with
+  `MODEL_UNAVAILABLE` naming it; pick another id or an
+  [API provider](/models).
 - **OpenCode Console** lists `go/` models only if you have Go. It leaves out
   models your workspace turned off, and Zen's free models. Key
   `providerOptions` by `opencode` for chat completions models, else by

package/docs/telemetry.mdx

@@ -34,10 +34,10 @@ export DO_NOT_TRACK=1                    # the cross-tool convention
 ```
 </CodeGroup>
 
-`e2e telemetry` prints the status and, when off, the reason.
-`e2e telemetry enable` reverses a saved `disable`. The environment variables
-win over the saved setting. The preferences file is under
-`$XDG_CONFIG_HOME/e2e/` when that is set, and `%APPDATA%\e2e\` on Windows.
+The environment variables win over the saved setting. `e2e telemetry enable`
+reverses a saved `disable`. `e2e telemetry` prints the status and, when off,
+the reason. The preferences file is under `$XDG_CONFIG_HOME/e2e/` when that is
+set, and `%APPDATA%\e2e\` on Windows.
 
 A `disable` also stops commands already running in another terminal. Nothing
 is sent when the preferences directory isn't writable, or from a source
@@ -67,13 +67,19 @@ sends nothing.
 Each command sends one `e2e_cli_session` event. `run` and `explore` also
 send `e2e_run_completed`, `init` sends `e2e_init_completed`, and `mcp` sends
 one `e2e_mcp_session` per MCP session. `--help` and `--version` send
-nothing, also on a command. `run()` from `e2e/runner` sends one
-`e2e_run_completed` per run that returns a report, and no session event. Its
-first call can print the notice below on stderr.
+nothing, on any command. `run()` from `e2e/runner` sends no session event,
+only one `e2e_run_completed` per run that returns a report.
 
-The events go in one request when the command ends, with a two-second
-deadline. `e2e mcp` sends each MCP session event in its own request. Events
-hold only the properties below.
+The events go in one request when the command or `run()` ends, with a
+two-second deadline. `e2e mcp` sends each session event in its own request,
+and `e2e run --watch` sends each run's event when that run ends.
+
+Events hold only the properties below. An error code that isn't an upper-case
+token is sent as `OTHER`. An engine, platform, model, fleet, or MCP client
+name, or an MCP client version, that isn't a plain token of letters, digits,
+dots, dashes, and underscores is sent as `other`. An engine version that isn't
+semver is sent as `unversioned`. In action and tool call counts, a project's
+own tools count together as `tool`, and any other name as `other`.
 
 <AccordionGroup>
 <Accordion title="Every event">
@@ -87,7 +93,7 @@ hold only the properties below.
 | `os`, `os_release`, `arch`, `cpu_count`, `memory_gb`, `docker`, `wsl`, `tty` | The machine class |
 | `sandbox` | The microVM the kernel announces itself as (`firecracker`), else `null` |
 | `ci`, `ci_name` | Whether `CI` is set, and the vendor when one announces itself (`github-actions`, `gitlab`, `circleci`, ...) |
-| `fleet` | The value of `E2E_TELEMETRY_FLEET` when it is a plain token, `other` when it isn't, `null` when unset |
+| `fleet` | The value of `E2E_TELEMETRY_FLEET`, `null` when unset |
 | `coding_agent` | The coding agent whose shell ran the command, when one announces itself (`claude-code`, `cursor`, `codex`, ...) |
 | `e2e_version`, `package_manager` | The e2e version, and the package manager that started the CLI |
 | `runtime`, `runtime_version`, `node_version` | `node`, `bun`, or `deno`, and its version. `node_version` is Node's own version, or the Node compatibility version under Bun or Deno |
@@ -101,7 +107,7 @@ hold only the properties below.
 | `command` | `init`, `run`, `list`, `cache ls`, `guide`, `telemetry`, ... |
 | `flags` | The names of the flags given, such as `--headed` or `--tag`. Never their values |
 | `exit_code`, `duration_ms` | How the command ended and how long it took |
-| `error_code` | The runner code of the failure that stopped the command before it could run (`CONFIG_NOT_FOUND`, `INVALID_CONFIG`, ...), `CLI_USAGE` for a flag or argument the CLI rejected, else `null`. A code that isn't an upper-case token is sent as `OTHER` |
+| `error_code` | The runner code of the failure that stopped the command before it could run (`CONFIG_NOT_FOUND`, `INVALID_CONFIG`, ...), `CLI_USAGE` for a flag or argument the CLI rejected, else `null` |
 
 </Accordion>
 <Accordion title="e2e_init_completed, once per e2e init">
@@ -123,16 +129,16 @@ hold only the properties below.
 | `flags` | The names of the flags given to the run command, never their values |
 | `tests_discovered`, `tests_selected`, `tests_executed`, `tests_passed`, `tests_failed`, `tests_interrupted`, `tests_flaky`, `tests_skipped` | The summary counts |
 | `attempts_total`, `tests_retried` | How many attempts the run made across tests and serial groups, and how many of these needed more than one |
-| `targets`, `platforms`, `engines` | How many targets, on which platforms, and with which engines and versions: `web`, `ios`, `web@0.6.1`, or your own engine's name and version. A name that isn't a plain token of letters, digits, dots, dashes, and underscores is sent as `other` |
+| `targets`, `platforms`, `engines` | How many targets, on which platforms, and with which engines and versions: `web`, `ios`, `web@0.6.1`, or your own engine's name and version |
 | `steps_total`, `steps_agent`, `steps_locator`, `steps_assertion`, `steps_screen`, `steps_app`, `steps_session`, `steps_resource` | Step counts by kind |
 | `steps_agent_act`, `steps_agent_assert`, `steps_agent_extract`, `steps_agent_waitfor` | The agent steps by entry point |
-| `tests_with_agent_steps`, `tests_with_act` | How many of the executed tests ran any agent step (the rest are deterministic end to end), and how many of those ran `agent.act` |
+| `tests_with_agent_steps`, `tests_with_act` | How many executed tests ran any agent step, and how many ran `agent.act` |
 | `agent_steps_replayed`, `agent_steps_partial`, `agent_steps_missed`, `agent_steps_vision` | How the replay cache served the agent steps, and how many used pixel evidence |
-| `agent_actions` | How often the agent took each action, by the runner's own names: `{"tap": 12, "typeText": 3, "scroll": 2}`. A project's own tools count together as `tool`, anything else as `other`. No other name is sent |
-| `model_gateway`, `model_provider`, `model_id`, `model_calls`, `model_tokens`, `model_cached_tokens`, `estimated_cost_usd` | For the first model-backed step: the AI SDK provider that served it (`gateway` for the Vercel AI Gateway, `openrouter.chat`, `openai.chat`, ...), the vendor (the first segment of a `vendor/model` id, else the provider), and the public model id (a fine-tuned or routed id is sent as `other`). Then the run's totals |
+| `agent_actions` | How often the agent took each action, by the runner's own names: `{"tap": 12, "typeText": 3, "scroll": 2}` |
+| `model_gateway`, `model_provider`, `model_id`, `model_calls`, `model_tokens`, `model_cached_tokens`, `estimated_cost_usd` | For the first model-backed step: the AI SDK provider that served it (`gateway` for the Vercel AI Gateway, `openrouter.chat`, `openai.chat`, ...), the vendor (the first segment of a `vendor/model` id, else the provider), and the public model id. Then the run's totals |
 | `cost_source` | `provider` when the provider priced the calls, `none` when nobody priced them, `null` for a run without model calls |
 | `artifact_bytes` | The size of the artifacts the run wrote |
-| `errors`, `error_codes` | How many run-level errors, and the distinct error codes seen anywhere in the run (`LOCATOR_NOT_FOUND`, `APP_UNREACHABLE`, ...), at most 20. A code that isn't an upper-case token is sent as `OTHER` |
+| `errors`, `error_codes` | How many run-level errors, and the distinct error codes seen anywhere in the run (`LOCATOR_NOT_FOUND`, `APP_UNREACHABLE`, ...), at most 20 |
 | `primary_error_code` | The first error the run recorded, which decided its status: a run-level error, else the failed attempt's verdict, else a step's. `null` for a clean run |
 | `error_kinds` | For `ENGINE_FAILURE`, `MODEL_PROVIDER_FAILED`, and `ERROR`: the code and a kind read off the message against a closed list (`auth`, `rate-limit`, `overloaded`, `timeout`, `network`, `server`, `bad-request`, `device`, `daemon`, `browser`, `permission`, `not-found`, `memory`, `other`), as `MODEL_PROVIDER_FAILED:rate-limit`. Only the kind's name leaves the machine, never the message |
 
@@ -168,13 +174,13 @@ client kills the server isn't sent.
 
 | Property | Value |
 | --- | --- |
-| `client`, `client_version` | The name and version the MCP client gives itself (`claude-code`, `cursor-vscode`, ...). A value that isn't a plain token is sent as `other`. `null` before the client introduced itself |
+| `client`, `client_version` | The name and version the MCP client gives itself (`claude-code`, `cursor-vscode`, ...). `null` before the client introduced itself |
 | `outcome`, `error_code` | `closed` for a session that opened, `open-failed` with the runner code of the failure (`APP_UNREACHABLE`, `ENGINE_FAILURE`, ...) |
 | `ended_by` | `agent` for `close_session`, `idle`, `step` for the session's time limit, `server` when the server shut down. `null` for a failed open |
 | `platform`, `engine`, `headed` | The target's platform and engine with its version, as for a run, and whether the UI was shown |
 | `duration_ms`, `sessions_open` | How long the session lived, and how many other sessions were open when it opened |
-| `calls_total`, `calls_failed`, `calls` | Tool calls, failed calls, and calls by the session's own tool names: `{"observe": 4, "tap": 9, "locate": 2, "screenshot": 1}`. A project's own tools count together as `tool`, anything else as `other`. No other name is sent |
-| `error_codes` | How often each runner code failed a call: `{"LOCATOR_NOT_FOUND": 2}`, at most 20 codes. A code that isn't an upper-case token is sent as `OTHER` |
+| `calls_total`, `calls_failed`, `calls` | Tool calls, failed calls, and calls by the session's own tool names: `{"observe": 4, "tap": 9, "locate": 2, "screenshot": 1}` |
+| `error_codes` | How often each runner code failed a call: `{"LOCATOR_NOT_FOUND": 2}`, at most 20 codes |
 
 </Accordion>
 </AccordionGroup>
@@ -217,25 +223,19 @@ other connection a run can make.
 
 ## Feedback
 
-[`e2e feedback`](/reference/cli#e2e-feedback) isn't telemetry. When you or
-your coding agent run it, it sends the words in its flags as one
-`e2e_feedback` event to the same PostHog project, with the machine facts
-above.
-
-- A saved `e2e telemetry disable` doesn't stop it.
-  `E2E_TELEMETRY_DISABLED=1` and `DO_NOT_TRACK=1` do.
-- `E2E_TELEMETRY_DEBUG=1` and `--dry-run` print the event instead of sending
-  it.
-- When telemetry is on, the event has the same anonymous id as your
-  telemetry events.
-- The CLI redacts well-known token shapes and the values of secret-named
-  environment variables, if 6 characters or longer, before it sends.
-- A bug without `--repro` isn't sent. You get a prefilled GitHub issue to
-  finish and submit yourself.
+[`e2e feedback`](/reference/cli#e2e-feedback) isn't telemetry. It sends the
+words in its flags as one `e2e_feedback` event to the same PostHog project,
+with the machine facts above. When telemetry is on, it has the same anonymous
+id as your telemetry events.
+
+A saved `e2e telemetry disable` doesn't stop it. `E2E_TELEMETRY_DISABLED=1`
+and `DO_NOT_TRACK=1` do. The [CLI reference](/reference/cli#e2e-feedback)
+covers redaction, `--dry-run`, and bugs without `--repro`.
 
 ## The notice
 
-The first command prints a notice on stderr that tells how to opt out, and
-leaves `--reporter json` output intact. `e2e init` and `e2e telemetry` don't
-print it, so it waits for the next command. The notice shows again after a
-large change to the data collected.
+The first command, or the first `run()` from `e2e/runner`, prints a notice
+on stderr that tells how to opt out. It leaves `--reporter json` output
+intact. `e2e init` and `e2e telemetry` don't print it, so it waits for the
+next command. The notice shows again after a large change to the data
+collected.

package/package.json

@@ -1,7 +1,8 @@
 {
   "name": "e2e",
-  "version": "0.21.1",
+  "version": "0.21.2",
   "description": "SDK, runner, and CLI for agentic end-to-end testing",
+  "mcpName": "io.github.tester-army/e2e",
   "keywords": [
     "e2e",
     "end-to-end",
@@ -132,7 +133,7 @@
     "@ai-sdk/openai": "4.0.90",
     "@ai-sdk/openai-compatible": "3.0.53",
     "@ai-sdk/xai": "5.0.4",
-    "@e2e-dev/web": "0.15.1",
+    "@e2e-dev/web": "0.15.2",
     "@modelcontextprotocol/client": "2.2.0",
     "@openrouter/ai-sdk-provider": "3.1.0",
     "@types/node": "26.6.2",

package/schema/report-v1.schema.json

@@ -233,6 +233,12 @@
       "required": ["mode", "replayedActions", "totalActions"],
       "properties": {
         "mode": { "enum": ["self-finalized", "agent-concluded", "missed"] },
+        "recordedBy": {
+          "type": "object",
+          "additionalProperties": false,
+          "required": ["provider", "model"],
+          "properties": { "provider": { "type": "string", "maxLength": 300, "pattern": "\\S" }, "model": { "type": "string", "maxLength": 300, "pattern": "\\S" } }
+        },
         "reason": {
           "enum": [
             "retry",

package/skills/e2e/references/agent.md

@@ -59,9 +59,11 @@ gates, and limits.
   for a custom brain; none inherits `model` or `context` from `default`.
 - `model` drives `agent.act`. Judgments use `judge` when set, else `model`.
 - A missing model for the built-in agent raises one run-level
-  `MODEL_UNAVAILABLE` when the first test acquires `agent`, exit 2; auth
-  failures surface on the first model call as `MODEL_PROVIDER_FAILED`, or as
-  `CONFIG_LOAD_FAILED` for a provider that reads its key when the config loads.
+  `MODEL_UNAVAILABLE` when the first test acquires `agent`, exit 2, and so
+  does a model GitHub Copilot does not serve to the account, at the first
+  call; auth failures surface on the first model call as
+  `MODEL_PROVIDER_FAILED`, or as `CONFIG_LOAD_FAILED` for a provider that
+  reads its key when the config loads.
 - `context` is what the app calls things, sent to every model call, judges
   included; `system` is how the acting agent works, read only by the act
   loop. Both live on the agents entry; `agentContext` on a test or group
@@ -279,6 +281,9 @@ selectors: the runner validates and authorizes every tool call first.
 
 ## The replay cache
 
+Replay reports expose the original model as `step.cache.recordedBy` when
+available; older recordings omit it. Use `--no-cache` to evaluate a new model.
+
 A passing `agent.act` saves its actions once a later check verifies the
 outcome; the next run replays them without model calls, and the live agent
 continues from the current screen when the app or final state no longer

package/skills/e2e/references/debugging.md

@@ -92,7 +92,7 @@ jq '.run.results[] | select(.selected and .status != "passed") | .attempts[-1]
 | `ACTION_FAILED` | Element not actionable (covered, disabled, detached), or an operation timed out | `expect` the condition first; close overlays; check `actionTimeout` |
 | `TEST_TIMEOUT` | The attempt exceeded `timeout` (120 s) | Split the test, or raise `timeout` |
 | `STEP_NOT_AWAITED` | The body returned while a step or `expect.poll` still ran, or an `afterEach` hook or fixture teardown returned while an `expect.poll` still ran: a call without `await` | `await` the call the code frame names (every `app`, `agent`, `screen`, `expect`, `expect.poll` call) |
-| `MODEL_UNAVAILABLE` | `agents.<name>.model` holds no AI SDK instance: reported once under `run.errors` at the first `agent` fixture; the run stops. A model-less custom executor fails only `waitFor` and `extract` this way | Construct one, e.g. `gateway('openai/gpt-6-luna-fast')` from `ai`, and export its key (`AI_GATEWAY_API_KEY`) |
+| `MODEL_UNAVAILABLE` | `agents.<name>.model` holds no AI SDK instance: reported once under `run.errors` at the first `agent` fixture; the run stops. A model-less custom executor fails only `waitFor` and `extract` this way. Also GitHub Copilot refusing the configured model to the account (`model_not_supported`, a plan or settings limit the listing cannot show): reported once at the first call, the run stops | Construct one, e.g. `gateway('openai/gpt-6-luna-fast')` from `ai`, and export its key (`AI_GATEWAY_API_KEY`); for Copilot, pick an id the plan serves |
 | `MODEL_PROVIDER_FAILED` | Network, 5xx, rate limit, no credits, or no response in 120 s, after the transport retries | Check the credential (the key; for keyless `gateway()` the Vercel CLI login and `.vercel/project.json`) and quota; retry (exit 3) |
 | `REPLAY_STALE` | `--strict-cache` or `cache.strict`, and a committed recording no longer replays; `step.cache.reason` says why | Rerun read-write with the knob the message names off; commit the changed entry under the directory it names (a custom `cache.store` is written by that run) |
 | `AUTOMATION_UNSUPPORTED` | The step needs an interaction the engine's toolset lacks (drag on a device, say); blocked, exit 1 | Do that step with `screen` actions, or run the test on a target that supports it |

package/skills/e2e/references/mcp.md

@@ -71,7 +71,7 @@ The catalog, per session:
 | Catalog tool | Does |
 | --- | --- |
 | `observe` | A fresh observation: one node per line as `#id role "name" ...`, plus the current path. On the web engine a link's `href` is its path on the app URL's origin, else origin and path (`mailto:` and `tel:` keep their scheme, `blob:` its inner origin and path); `?…` and `#…` mark a dropped query or fragment, `data:…` and `javascript:…` an inline payload, a trailing `…` a target cut at 256 characters. |
-| `tap`, `double_tap`, `long_press`, `right_click`, `hover`, `type`, `press`, `select`, `check`, `scroll`, `scroll_to`, `drag`, `upload`, `navigate`, `back`, `dismiss_keyboard` | The grammar verbs, as the testing agent gets them: `check` takes `checked`, `drag` a `to` node, `upload` project-relative `files`; `dismiss_keyboard` comes with device engines. Each reports what changed on screen; `observe` shows the whole screen. A verb the engine cannot honor is not listed and fails with `UNSUPPORTED_CAPABILITY`. |
+| `tap`, `double_tap`, `long_press`, `right_click`, `hover`, `type`, `press`, `select`, `check`, `scroll`, `scroll_to`, `drag`, `upload`, `navigate`, `back`, `dismiss_keyboard` | The grammar verbs, as the testing agent gets them: `check` takes `checked`, `drag` a `to` node, `upload` project-relative `files`; `dismiss_keyboard` comes with device engines. On an engine that declares `tapModifiers`, `tap`, `double_tap`, and `right_click` take an optional `modifiers` list (`Shift`, `Control`, `Alt`, `Meta`, `ControlOrMeta`) to hold keys for the click. Each reports what changed on screen; `observe` shows the whole screen. A verb the engine cannot honor is not listed and fails with `UNSUPPORTED_CAPABILITY`. |
 | `type_secret` | Fills a configured secret by name: a credential's password into a password field, a `secrets` entry into any editable input; the plaintext never reaches the agent. Listed when the config declares `credentials` or `secrets` and the engine can fill secrets. |
 | `locate` | Tries a semantic locator with exactly one of `role`, `text`, `label`, `placeholder`, `testId` (`name` only with `role`; `exact` optional; anything else is `INVALID_ARGUMENT`) and returns how many nodes match, which, and the `screen.*` call to write. |
 | `screenshot` | The masked pixels as an image. |

package/skills/e2e/references/setup.md

@@ -90,7 +90,10 @@ Switching to Copilot: install `ai`, `@ai-sdk/openai-compatible`, and
 over Copilot's Responses API when the plan serves that model only there, choosing
 per model from the plan's listing. `npx e2e models github-copilot` marks the models
 it cannot call at all: those served only over an API `copilot()` does not speak, and
-those the plan has not enabled.
+those the plan has not enabled. Whether the plan serves a model shows nowhere: on
+Copilot Free, GitHub serves models through auto model selection only, so a model
+called by id is usually refused, and the run stops with `MODEL_UNAVAILABLE` naming
+it. Pick another id or an API provider.
 
 Switching to OpenCode Console: install `ai`, `@ai-sdk/openai-compatible`,
 `@ai-sdk/openai`, `@ai-sdk/anthropic`, and `@ai-sdk/google`, set
@@ -201,9 +204,9 @@ start a script that brings them up and serves the app.
 
 | Option | Meaning |
 | --- | --- |
-| `browser` | `'chromium'` (default), `'firefox'`, `'webkit'`, or a `BrowserProvider` leasing hosted browsers over CDP (`kernel()` from `@e2e-dev/kernel`, `smol()` from `@e2e-dev/smol`, or your own), which implies chromium and excludes `connect`. Scope `'worker'` (default): one browser per worker slot from `prepare` to `finish`; `'attempt'`: one per attempt, with `reconnectEndpoint`'s limits. |
+| `browser` | `'chromium'` (default), `'firefox'`, `'webkit'`, or a `BrowserProvider` leasing hosted browsers over CDP (`kernel()` from `@e2e-dev/kernel`, `smol()` from `@e2e-dev/smol`, `lightpanda()` from `@e2e-dev/lightpanda`, or your own), which implies chromium and excludes `connect`. Scope `'worker'` (default): one browser per worker slot from `prepare` to `finish`; `'attempt'`: one per attempt, with `reconnectEndpoint`'s limits (`lightpanda()` is worker scope only). |
 | `viewport` | `{ width, height }`, default 1280x720; `null` follows the browser window (hosted live view, headed run). On a headed hosted browser (Kernel) use `null` and size the service's screen; a fixed size gives a smaller, unmaximized window. |
-| `connect` | `{ cdpEndpoint }` attaches to a remote Chromium over CDP; both it and `reconnectEndpoint` are resolvers `(signal) => url`, not strings. With `reconnectEndpoint` it rides one persistent default context and reconnects only to the original browser and page. |
+| `connect` | `{ cdpEndpoint }` attaches to a remote Chromium over CDP; `cdpEndpoint` is required, and both it and `reconnectEndpoint` are resolvers `(signal) => url`, not strings, else `INVALID_CONFIG` at load. With `reconnectEndpoint` it rides one persistent default context, needs a fresh browser from `cdpEndpoint` every attempt, and reconnects only to the original browser and page. |
 | `headers` | Sent to the app's site only (Vercel's `x-vercel-protection-bypass`, ngrok's `ngrok-skip-browser-warning`), `agent.act` included; disables the browser HTTP cache and service workers. |
 | `basicAuth` | `{ username, password }` for a `401` challenge; `password` may be `secrets.get('name')`, resolved per attempt and redacted like any secret, the base64 `Authorization` credential too. |
 | `userAgent` | The `User-Agent` every attempt sends and `navigator.userAgent` reports. |
@@ -361,6 +364,12 @@ export default {
   iOS keyboard actions such as `locator.press('Enter')` return
   `UNSUPPORTED_CAPABILITY`; tap an app button when available.
   See `examples/with-limrun` for the same form and restart tests on both platforms.
+- `testmu({ device, osVersion, app })` from `@e2e-dev/testmu` leases TestMu
+  AI (formerly LambdaTest) emulators, simulators, or real devices
+  (`deviceType: 'real'`); `device` and `osVersion` match its catalog
+  exactly, TestMu AI installs `app` (omit `app.appPath`), and it reads
+  `LT_USERNAME` and `LT_ACCESS_KEY`. An attempt's video links TestMu AI's
+  recording of the whole session.
 - Only a control that appeared or moved with the previous action waits out
   `transition` (default 500 ms); agent actions settle `settle` ms (default
   150) before the next observation, `settle: false` skips it.

package/skills/e2e/references/writing-tests.md

@@ -144,7 +144,7 @@ assertion. Every query also exists on a locator, scoped to its subtree.
 | `getByPlaceholder(text, { exact?, visible? })` | Inputs by placeholder. |
 | `getByText(text, { exact?, visible? })` | Visible text. |
 | `getByDisplayValue(value, { exact?, visible? })` | Inputs by current value; on the web it cannot scope child queries or be a `filter({ has })` target. |
-| `getByTestId(id, { visible? })` | `data-testid` on the web (or `web({ testIdAttribute })`), accessibility identifier or resource id on a device (React Native `testID`, SwiftUI `.accessibilityIdentifier`, Compose `testTag` with `testTagsAsResourceId` on Android, Flutter `Semantics(identifier:)`); a string matches the whole id, a RegExp tests it. Last resort. |
+| `getByTestId(id, { visible? })` | `data-testid` on the web (or `web({ testIdAttribute })`), accessibility identifier or resource id on a device (React Native `testID`, SwiftUI `.accessibilityIdentifier`, Compose `testTag` with `testTagsAsResourceId` on Android, Flutter `Semantics(identifier:)`, .NET MAUI `AutomationId`); a string matches the whole id, a RegExp tests it; on Android the `<package>:id/` prefix may be left off. Last resort. |
 
 Roles: `button`, `link`, `textbox`, `searchbox`, `combobox`, `listbox`,
 `option`, `checkbox`, `radio`, `radiogroup`, `switch`, `slider`, `spinbutton`,