version: "2.1.0"
date: "2026-10-08"
ego-browser
Call help() or help("page.click") to print signatures and uncommon options
of APIs named below; it prints by itself, so do not wrap it in console.log.
For installation or connection problems, read references/install.md in the
ego-browser Skill directory.
Run browser scripts
Pass the script to ego-browser nodejs. The form depends on the shell that runs
the command: heredoc is Bash/Zsh syntax and does not exist in PowerShell or cmd.
Bash, Zsh, Git Bash
ego-browser nodejs <<'EOF'
const task = await taskSpace("inspect example page");
const page = task.page("p1");
await page.goto("https://example.com");
console.log({ taskSpaceId: task.spaceId, page: page.label });
console.log(await page.snapshot());
EOF
When a sandbox blocks heredoc input, use -e instead. Use single quotes around
the code and double quotes for JavaScript strings; single quotes within the code
require shell quoting.
ego-browser nodejs -e '
const task = await taskSpace("inspect example page");
await task.page("p1").goto("https://example.com");
console.log(task.spaceId);
'
PowerShell
PowerShell has no heredoc, and Windows PowerShell 5.1 drops the double quotes
inside an argument when it calls a native program, which splits the script into
several arguments. Do not pass a script that contains double quotes or spans
several lines to -e. Write it to a UTF-8 .mjs file instead and evaluate an
ASCII-only one-line entry point, which needs no double quotes of its own:
ego-browser nodejs -e "await import('file:///C:/Users/<name>/ego/task.mjs')"
The imported module runs with the same helpers and console as an inline
script, so the whole task belongs in the .mjs file. Use forward slashes in the
file:/// URL. Non-ASCII content such as Chinese text must stay inside the
file: never put it in a -e argument.
Do not pipe the file in with Get-Content. Windows PowerShell 5.1 reads a UTF-8
file as ANSI and sends ASCII to native programs, which corrupts non-ASCII text
and truncates strings. PowerShell has no < input redirection either.
cmd
cmd has no heredoc, but it does support input redirection, so it can run the
same .mjs file directly:
ego-browser nodejs < C:\Users\<name>\ego\task.mjs
Later rounds with a script file
Each run is a new process, and earlier rounds have already run. Write a new,
complete file for every round (round-01.mjs, round-02.mjs, ...) instead of
editing the previous one, which would replay old steps. Keep one goal's files in
a new directory named for the goal plus the date and time, such as
C:/Users/<name>/ego/laptop-prices-20260922-1405/, and only run files written
in this conversation. Create files with your file tool, not PowerShell
Set-Content, Out-File, or >, which write ANSI or UTF-16 on 5.1.
When the script fails to parse
SyntaxError: await is only valid in async functions and the top level bodies of modules almost always means the shell damaged the script before Node received
it; top-level await itself is supported. Do not wrap the code in an async
IIFE, open a new task space, or resend the same command. Switch to the .mjs
file form above.
The script always runs in Node.js, not in the web Page. Browser helpers and
Node.js APIs belong in the script; Page globals such as window, document,
location, and DOM APIs do not. Put browser-side JavaScript inside
page.evaluate(). Do not import Playwright or launch another browser.
The Node.js runtime uses ESM. When a script needs local files, load built-ins
with dynamic imports such as await import("node:fs/promises"). The script
does not run in your shell's working directory, so use absolute paths for every
file it reads or writes. It also does not inherit your shell's environment
variables: process.env will not contain values you set before running
ego-browser, so write every input directly into the script.
Ego-browser deliberately exposes a small custom API. It is not Playwright, even
where method names and options look similar. Use only the TaskSpace, Page,
FileChooser, mouse, and keyboard APIs explicitly listed in this Skill. Do not
infer Playwright methods such as locator(), getByRole(), context(),
expect(), or route(). When the listed API does not cover an operation, use
the documented page.evaluate() or page.cdp() escape hatches instead of
guessing another method.
Only these pointer actions accept an optional label option: click(),
dblclick(), hover(), dragAndDrop(), mouse.click(), mouse.move(), and
mouse.wheel(). Pass a concise 3-6 word description to keep the action text
next to the visible agent cursor in sync with the action. Other methods, such
as fill(), press(), focus(), selectOption(), waitForSelector(),
keyboard.*(), mouse.down(), and mouse.up(), reject a label option.
The label in selectOption(selector, { label }) is different: it selects the
option whose visible text matches.
When the user explicitly asks for ego-browser, start with a real browser command
and diagnose the CLI or installation only if it fails.
Spaces, rounds, and pages
- Use exactly one TaskSpace for the entire user goal. Create it once, print its
spaceId, and resume that same space in later rounds. Use multiple spaces
only when the user explicitly requests them.
- Task space names are unique.
taskSpace(name) resumes your own space with
that name instead of creating a second one. If the name belongs to a
user-owned space, it fails with EGO_TASK_SPACE_NAME_IN_USE: ask the user
whether to work in that space (then claimTaskSpace(id)), or pick another
name.
- Never use a new TaskSpace to recover from a stuck, blocked, timed-out, or
unexpected Page. Recover within the existing space; if it cannot continue,
stop and ask the user.
- Every invocation starts a new Node.js process. Task spaces, tabs, and Page labels
persist; JavaScript variables do not.
- A new task space starts with Page
p1; navigate it instead of opening
another Page.
- Reuse a Page with
goto() instead of opening a new Page for every URL.
- All time values are milliseconds.
// Later round: use the space id and Page label printed earlier.
const resumed = await taskSpace(7);
const source = resumed.page("p1");
await source.goto("https://example.com/releases");
Do not inspect or select profiles unless the user explicitly requests a
particular Ego Lite profile. A profileId applies only when creating a space;
use help("profiles") for the exact workflow.
Supported TaskSpace API:
- State:
spaceId, name, ownership, page(label), userPage()
- Pages:
await task.pages(), await task.tabs(), newPage(),
adopt(page, { as? }), release(label)
- Control:
waitForControl(options), handOff(), finish({ keep })
- Advanced:
cdp(method, params, options)
Pages receive permanent labels such as p1, p2, and p3. Prefer these labels
to custom { as } values. Reuse or close Pages as the task proceeds; the runtime
reports the configured Page budget when it is reached.
task.newPage() creates another blank Page when multiple Pages must stay open.
Navigate it separately with page.goto().
task.page(label) only refers to a Page that already exists; it never creates
one, and its handle is always truthy, so task.page("p2") || await task.newPage()
and if (!page) do not detect a missing label. Use labels printed by earlier
output or by await task.pages(), and call await task.newPage() to create one.
await task.pages() returns managed Pages. await task.tabs() returns every tab in the
space as { label?, page, targetId, title, url, active, openedBy }. A tab
without a label is unmanaged; adopt it before operating:
const active = (await task.tabs()).find((item) => item.active);
if (active && !active.label) {
const page = await task.adopt(active.page);
console.log({ page: page.label, url: await page.url() });
}
release(label) returns an unknown-origin Page to the user without closing its
tab. Close Agent-created Pages with page.close(). Treat openedBy: "unknown"
as user-owned when deciding whether a Page may be closed.
Page operations
ego-browser provides the following Page API:
- State and observation:
label, spaceId, openedBy, targetId, url(),
title(), info(), snapshot(), screenshot()
- Navigation and waits:
goto(), reload(), waitForURL(),
waitForEvent(), waitForSelector(), waitForLoadState(),
waitForFunction(), waitForTimeout()
- Elements:
click(), dblclick(), hover(), dragAndDrop(), fill(),
selectOption(), focus(), press(), setInputFiles(),
waitForFileChooser(), close()
- Dialogs:
acceptDialog(promptText?), dismissDialog()
- Pointer:
mouse.click(), move(), down(), up(), wheel()
- Keyboard:
keyboard.down(), up(), press(), type(), insertText(),
paste()
- Page code and protocols:
evaluate(fnOrString, argument),
fetch(url, options), cdp(method, params, options)
page.evaluate() callbacks run only inside the Page; they cannot read variables
or Node.js modules from the surrounding script. Define browser-side helpers
inside the callback or pass one JSON-serializable value as its second argument.
Work efficiently:
- Each time you observe, collect only the cheapest page state sufficient to
choose the next action. Use a snapshot for semantic or locator ground truth
and a screenshot for visual confirmation; do not request both by default.
- If an action does not produce the expected result, inspect the current page
before deciding whether to retry. Do not blindly repeat it or immediately
fall back to coordinates or raw CDP.
- Once the page clearly shows the requested result, stop; do not confirm the
same result through multiple surfaces.
Semantic pages: snapshot and selectors
Prefer snapshots and semantic selectors for ordinary DOM pages. Use screenshots
and coordinates only when useful DOM semantics are unavailable.
Before choosing an unfamiliar target, take a snapshot. When the current state
is sufficient to plan several actions on the same Page, complete them in one
script invocation, then observe the result once. Observe between actions only when an
intermediate result changes what should happen next. Keep the action sequence,
the wait for its final expected state, and the next snapshot in the same
script invocation. Print the snapshot last so the next round can act on it directly.
The final snapshot is the next round's starting view of the changed page;
without it, that round usually has to spend a separate browser call observing
before it can choose the next target, which wastes compute.
Wait for the expected result: use waitForURL() for navigation,
waitForSelector() for element state, or waitForFunction() for application
state. Avoid fixed delays when an observable condition exists. A snapshot
captures the current moment; it does not wait for the page to become stable.
page.snapshot() captures the current viewport. For content outside it, use
page.snapshot({ scope: "full_page" }).
The default viewport snapshot includes visible iframe content returned by the
browser. To focus on a frame's subtree, reuse the ref printed on its iframe
line:
console.log(await page.snapshot({ scope: "subtree", root: "@12" }));
Use the refs returned by the subtree for actions inside the iframe. A subtree
snapshot does not scope later locator actions; they still prefer actionable
matches in the top document before searching frames.
waitForLoadState() defaults to load. waitForFunction() follows the
Playwright argument order; pass undefined before options when there is no Page
argument:
await page.waitForFunction(() => window.appReady, undefined, {
timeout: 10_000,
});
// Round 1: inspect and choose targets from this output.
const page = task.page("p1");
console.log(await page.snapshot());
// Next round: act using the previous output, verify, then prepare the next round.
const page = task.page("p1");
await page.fill("@21", "[email protected]");
await page.click("loc=role:button[name='Sign in']");
await page.waitForSelector("loc=css:#account-home", { state: "visible" });
console.log(await page.snapshot());
Element actions accept:
- snapshot refs such as
@21 or ref=21
text=... for page contentloc=css:, loc=role:, and loc=href: locatorsxpath=...- raw CSS selectors
Selector actions require exactly one match. Unquoted text normalizes whitespace,
ignores case, and matches a substring; quoted text such as
text="Save changes" is exact and case-sensitive.
A small Playwright-compatible selector subset is also accepted: css=...,
terminal :has-text("...") and :text-is("..."), >> nth=N after a CSS,
text, or href selector (N is -1 or non-negative), plus
loc=role:...[name*="..."] for accessible-name substrings. Other Playwright
selector syntax is not supported.
When a selector identifies a wrapper, focus() and press() may use its
interactive ancestor or unique editable descendant; fill() and
setInputFiles() only continue to a unique compatible control.
click(), fill(), hover(), and dragAndDrop() automatically bring their
target into view with browser wheel input. Do not pre-scroll solely to make a
DOM target actionable.
Snapshot node names are accessibility roles. Use a ref now or loc=... to find
the element again. After the page changes, take a new snapshot. When a useful
node has no ref, construct a selector from its role, text, or surrounding
context. CSS searches nested open shadow roots. Actions use an actionable match
in the top document first, then search frames when the top document has none.
Multiple actionable matches in the selected document or frame are ambiguous.
Select options by value, visible label, or zero-based index. A string matches
either value or label; pass an array for a multiple select:
await page.selectOption("select[name=month]", { label: "October" });
Pass null or [] to clear the current selection.
Visual pages: screenshot, mouse, and keyboard
Use a screenshot with mouse and keyboard operations for canvas, rich-text,
spreadsheets, maps, and other interfaces that lack useful DOM semantics:
const path = await page.screenshot({ path: "/absolute/path/before.png" });
await page.mouse.click(420, 260, { label: "open spreadsheet cell" });
await page.mouse.wheel(0, 600, { label: "scroll project board" });
await page.keyboard.paste("hello\tworld");
console.log({ screenshot: path });
path must be absolute; relative paths are rejected. Omit path to write
into the system temporary directory and use the returned file path. On
Windows, write paths with forward slashes, such as
C:/Users/<name>/ego-shots/before.png, because backslashes are escape
characters inside JavaScript strings.
Inspect the screenshot with an image-viewing tool. Coordinates use CSS pixels;
keyboard names and +-separated chords follow Playwright syntax. Use
ControlOrMeta for portable shortcuts and verify the resulting page state.
mouse.wheel() performs a short wheel-input motion at the current mouse
position and resolves when that motion completes. In each script invocation, move or
click over the intended scrollable area before using it.
On macOS and Windows, keyboard.paste() sends the native paste shortcut and
then restores the user's clipboard. Pass { text, html } when a rich editor
needs structured clipboard content; text is the plain-text fallback. On other
platforms, use keyboard.insertText() for plain text.
await page.keyboard.paste({
text: "Name\tStatus",
html: "<table><tr><td>Name</td><td>Status</td></tr></table>",
});
For rich-text editors and editable grids, validate a small edit before repeating
it at scale. Canvas-backed editors may not expose visible content through DOM
text or selectors; verify those results with a screenshot or an
application-specific visible state.
Page JavaScript and CDP
Use page.evaluate() for bulk extraction or complex in-page work. It accepts
one JSON-serializable argument and returns a JSON-serializable value:
const rows = await page.evaluate(
({ selector, limit }) =>
[...document.querySelectorAll(selector)].slice(0, limit).map((node) => ({
text: node.textContent?.trim(),
href: node.querySelector("a")?.href,
})),
{ selector: "article", limit: 20 },
);
page.evaluate() has no timeout option. Keep long work in bounded calls; on a
safety timeout, use executionStopped and mayHaveLateEffects to decide
whether an unsafe follow-up requires reloading or closing the Page first.
Use documented Page methods first. If a wrapper is missing or does not work
reliably on the current page, use page.cdp() as a lower-level path for
diagnosis or control. It accepts Page, Runtime, DOM, Network, Input, and similar
commands; use task.cdp() for Target and Browser commands. Raw CDP invalidates
refs. Do not persist page.targetId across rounds.
Action receipts, popups, and dialogs
When an action is expected to open a new Page, start the wait before the action:
const popupPromise = page.waitForEvent("popup");
await page.click('a[target="_blank"]');
const popupPage = await popupPromise;
await popupPage.waitForLoadState();
High-level actions also report immediately observed popups in receipt.popups
as { label, targetId }. Resolve the Page with
task.page(receipt.popups[0].label) and continue there; wait for its URL when
the destination matters.
For uncommon protocol-event workflows, await page.events() returns and clears
the buffered event array; it is not an EventEmitter.
A synchronous JavaScript dialog may appear as receipt.dialog or in
page.info(). Handle it before continuing:
await page.acceptDialog("prompt response");
// Or: await page.dismissDialog();
A receipt describes only the dispatched action and immediate popup or dialog
observations; it does not verify the resulting application state.
Files and requests
Set an existing file input with absolute paths:
await page.setInputFiles("input[type=file]", ["/absolute/path/report.pdf"]);
If a click creates the file input, start waiting before the click:
const chooserPromise = page.waitForFileChooser({ timeout: 10_000 });
await page.click("button.upload");
const chooser = await chooserPromise;
const result = await chooser.setFiles("/absolute/path/report.pdf");
An upload-triggered JavaScript dialog may be returned as result.dialog; when
present, handle it with the dialog methods above.
For a browser download, arm the event before the triggering action and save the
returned artifact to an absolute path in the same script:
const downloadPromise = page.waitForEvent("download", { timeout: 30_000 });
await page.click("button.download");
const download = await downloadPromise;
console.log({
url: download.url(),
suggestedFilename: download.suggestedFilename(),
});
await download.saveAs("/absolute/path/report.pdf");
download.saveAs() waits for completion and creates missing parent
directories. download.path() returns the round-local temporary file;
failure(), cancel(), and delete() manage its lifecycle. Temporary download
files are removed when the SDK round is disposed, so call saveAs() before the
script ends. Do not set a global download directory with raw CDP; each download
wait configures and restores only the addressed Page session.
page.fetch() runs window.fetch() in the Page: relative URLs, cookies, and
service workers use that Page, and browser CORS still applies. It returns
{ ok, status, statusText, url, headers, body }:
const response = await page.fetch("/api/items", {
method: "POST",
headers: { "content-type": "application/json" },
body: JSON.stringify({ limit: 20 }),
timeout: 10_000,
});
Save binary responses without converting them to text:
await page.fetch("/image.png", { saveAs: "/absolute/path/image.png" });
Use standard Node.js fetch() for background requests that do not need Page
browser semantics.
User control and completion
Stop when the user takes control or the space is inactive or unassigned. Do not
retry or route around the stop. Permission prompts, device choosers, and
other browser-owned prompts require the user to handle them.
When the user must act in the browser, call await task.handOff(), end the
round, and explain what they should do. After the user confirms, resume the
same space:
const task = await takeOverTaskSpace(7);
const userPage = task.userPage();
Adopt userPage if it is unmanaged. Use waitForControl() only when the current
script must wait in place. Claim a user-owned or inactive space only when the
user explicitly asks; this includes a space you finished with kept Pages, which
takeOverTaskSpace() cannot resume. Find its numeric id first; names may be
duplicated. listTaskSpaces() entries expose that id as id (spaceId exists
only on a TaskSpace handle), so pass entry.id to claimTaskSpace() or
taskSpace():
const spaces = await listTaskSpaces();
console.log(spaces.filter((space) => space.ownership === "user"));
const task = await claimTaskSpace(7);
const userPage = task.userPage();
When the task succeeds, close the TaskSpace by default with
await task.finish({ keep: [] }). Call finish() exactly once and wait for it
to resolve before reporting completion.
finish() ends the task and all of its Page labels, including Pages kept for the
user. Put it last in the final round only; do not operate any Page after it in
the same script or in later rounds.
Keeping Pages is a rare exception: retain only necessary Pages when the user
explicitly asks, or when the result must remain in the browser for the user to
view or continue working with. Pages merely visited, search results, and
intermediate steps do not need to remain open.
await task.finish({ keep: [] }); // Default: keep no Agent-managed Pages.
await task.finish({ keep: ["p2"] }); // Exception: keep only the result Page for the user.
User-created and unmanaged tabs are protected; if any remain, keep: [] does
not close the whole space. Do not close unwanted Pages one by one at completion;
list the Pages to keep instead.
Use page.close() only while the task is still in progress. Do not call
finish() when the task stops for user control or an error.
If the final output contains [ego-browser:notice], finish the current browser
task, tell the user an Ego Lite update is available, and run
ego-browser upgrade only with their approval. Run ego-browser skill again
after the upgrade.
Clearing cookies, cache, and storage
Before clearing any cookie, cache, or storage, read
references/clearing-state.md in the ego-browser Skill directory. Some clears
reach the whole browser profile.