core

Name: core
Author: vercel-labs

39kstars2.5kforks

How to install core?

vercel-labs/agent-browser/core

$npx -y skills add vercel-labs/agent-browser --skill core

Installs into the current project.

Summary

Core agent-browser usage guide. Read this before running any agent-browser commands. Covers the snapshot-and-ref workflow, navigating pages, interacting with elements (click, fill, type, select), extracting text and data, taking screenshots, managing tabs, handling forms and auth

Files · 1

View on GitHub

SKILL.md

1	# agent-browser core
2
3	Fast browser automation CLI for AI agents. Chrome/Chromium via CDP, no Playwright or Puppeteer dependency. Accessibility-tree snapshots with compact `@eN` refs let agents interact with pages in ~200-400 tokens instead of parsing raw HTML.
4
5	Most normal web tasks (navigate, read, click, fill, extract, screenshot) are covered here. Load a specialized skill when the task falls outside browser web pages — see [When to load another skill](#when-to-load-another-skill).
6
7	## The core loop
8
9	```bash
10	agent-browser open <url> # 1. Open a page
11	agent-browser snapshot -i # 2. See what's on it (interactive elements only)
12	agent-browser click @e3 # 3. Act on refs from the snapshot
13	agent-browser snapshot -i # 4. Re-snapshot after any page change
14	```
15
16	Refs (`@e1`, `@e2`, ...) are assigned fresh on every snapshot. They become stale the moment the page changes — after clicks that navigate, form submits, dynamic re-renders, dialog opens. Always re-snapshot before your next ref interaction.
17
18	## Quickstart
19
20	```bash
21	# Install once
22	npm i -g agent-browser && agent-browser install
23
24	# Linux hosts can install required browser libraries too
25	agent-browser install --with-deps
26
27	# Take a screenshot of a page
28	agent-browser open https://example.com
29	agent-browser screenshot home.png
30	agent-browser close
31
32	# Search, click a result, and capture it
33	agent-browser open https://duckduckgo.com
34	agent-browser snapshot -i # find the search box ref
35	agent-browser fill @e1 "agent-browser cli"
36	agent-browser press Enter
37	agent-browser wait --load networkidle
38	agent-browser snapshot -i # refs now reflect results
39	agent-browser click @e5 # click a result
40	agent-browser screenshot result.png
41	```
42
43	The browser stays running across commands so these feel like a single session. Use `agent-browser close` (or `close --all`) when you're done.
44
45	## MCP integration
46
47	For tools that support Model Context Protocol servers, start the stdio server:
48
49	```bash
50	agent-browser mcp
51	agent-browser mcp --tools all
52	agent-browser mcp --tools core,network,react
53	```
54
55	Configure the MCP client to launch `agent-browser` with `["mcp"]`. The server defaults to MCP protocol 2025-11-25 and accepts older supported client protocol versions during initialization. The default tools profile is `core`, which keeps MCP context small for everyday browser automation. Use `--tools all` for the full typed CLI parity surface, or combine profiles with commas, such as `--tools core,network,react`. Profiles are `core`, `network`, `state`, `debug`, `tabs`, `react`, `mobile`, and `all`; the `debug` profile includes plugin registry and command.run tools. Each tool accepts typed arguments plus `extraArgs` for advanced CLI flags and exact CLI parity. The common `allowedDomains` array maps to `--allowed-domains` and activates the same WebRTC containment and launch-mode restrictions. Tool discovery is paginated and includes read-only/open-world annotations so modern MCP clients can load the large typed surface incrementally. Use the tool `session` argument or `AGENT_BROWSER_SESSION` to isolate browser sessions.
56
57	## eve agent integration
58
59	For eve agents, mount the `@agent-browser/eve` extension instead of hand-writing browser tools. It adds namespaced tools such as `browser__navigate`, `browser__snapshot`, `browser__click`, `browser__fill`, `browser__find`, and `browser__screenshot`, all backed by agent-browser running inside the eve sandbox. The sandbox bootstrap helpers (`installAgentBrowser`, `agentBrowserRevalidationKey`) ship with the same package under `@agent-browser/eve/sandbox`, so `agent/sandbox.ts` needs no extra dependency.
60
61	## Reading a page
62
63	```bash
64	agent-browser snapshot # full tree (verbose)
65	agent-browser snapshot -i # interactive elements only (preferred)
66	agent-browser snapshot -i -u # include href urls on links
67	agent-browser snapshot -i -c # compact (no empty structural nodes)
68	agent-browser snapshot -i -d 3 # cap depth at 3 levels
69	agent-browser snapshot -s "#main" # scope to a CSS selector
70	agent-browser snapshot -i --json # machine-readable output
71	```
72
73	Snapshot output looks like:
74
75	```
76	Page: Example - Log in
77	URL: https://example.com/login
78
79	@e1 [heading] "Log i

People who liked this also viewed

6 picks

Type

core

39kstars2.5kforks

View on GitHub

How to install core?

vercel-labs/agent-browser/core

$npx -y skills add vercel-labs/agent-browser --skill core

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# agent-browser core
2
3	Fast browser automation CLI for AI agents. Chrome/Chromium via CDP, no Playwright or Puppeteer dependency. Accessibility-tree snapshots with compact `@eN` refs let agents interact with pages in ~200-400 tokens instead of parsing raw HTML.
4
5	Most normal web tasks (navigate, read, click, fill, extract, screenshot) are covered here. Load a specialized skill when the task falls outside browser web pages — see [When to load another skill](#when-to-load-another-skill).
6
7	## The core loop
8
9	```bash
10	agent-browser open <url> # 1. Open a page
11	agent-browser snapshot -i # 2. See what's on it (interactive elements only)
12	agent-browser click @e3 # 3. Act on refs from the snapshot
13	agent-browser snapshot -i # 4. Re-snapshot after any page change
14	```
15
16	Refs (`@e1`, `@e2`, ...) are assigned fresh on every snapshot. They become stale the moment the page changes — after clicks that navigate, form submits, dynamic re-renders, dialog opens. Always re-snapshot before your next ref interaction.
17
18	## Quickstart
19
20	```bash
21	# Install once
22	npm i -g agent-browser && agent-browser install
23
24	# Linux hosts can install required browser libraries too
25	agent-browser install --with-deps
26
27	# Take a screenshot of a page
28	agent-browser open https://example.com
29	agent-browser screenshot home.png
30	agent-browser close
31
32	# Search, click a result, and capture it
33	agent-browser open https://duckduckgo.com
34	agent-browser snapshot -i # find the search box ref
35	agent-browser fill @e1 "agent-browser cli"
36	agent-browser press Enter
37	agent-browser wait --load networkidle
38	agent-browser snapshot -i # refs now reflect results
39	agent-browser click @e5 # click a result
40	agent-browser screenshot result.png
41	```
42
43	The browser stays running across commands so these feel like a single session. Use `agent-browser close` (or `close --all`) when you're done.
44
45	## MCP integration
46
47	For tools that support Model Context Protocol servers, start the stdio server:
48
49	```bash
50	agent-browser mcp
51	agent-browser mcp --tools all
52	agent-browser mcp --tools core,network,react
53	```
54
55	Configure the MCP client to launch `agent-browser` with `["mcp"]`. The server defaults to MCP protocol 2025-11-25 and accepts older supported client protocol versions during initialization. The default tools profile is `core`, which keeps MCP context small for everyday browser automation. Use `--tools all` for the full typed CLI parity surface, or combine profiles with commas, such as `--tools core,network,react`. Profiles are `core`, `network`, `state`, `debug`, `tabs`, `react`, `mobile`, and `all`; the `debug` profile includes plugin registry and command.run tools. Each tool accepts typed arguments plus `extraArgs` for advanced CLI flags and exact CLI parity. The common `allowedDomains` array maps to `--allowed-domains` and activates the same WebRTC containment and launch-mode restrictions. Tool discovery is paginated and includes read-only/open-world annotations so modern MCP clients can load the large typed surface incrementally. Use the tool `session` argument or `AGENT_BROWSER_SESSION` to isolate browser sessions.
56
57	## eve agent integration
58
59	For eve agents, mount the `@agent-browser/eve` extension instead of hand-writing browser tools. It adds namespaced tools such as `browser__navigate`, `browser__snapshot`, `browser__click`, `browser__fill`, `browser__find`, and `browser__screenshot`, all backed by agent-browser running inside the eve sandbox. The sandbox bootstrap helpers (`installAgentBrowser`, `agentBrowserRevalidationKey`) ship with the same package under `@agent-browser/eve/sandbox`, so `agent/sandbox.ts` needs no extra dependency.
60
61	## Reading a page
62
63	```bash
64	agent-browser snapshot # full tree (verbose)
65	agent-browser snapshot -i # interactive elements only (preferred)
66	agent-browser snapshot -i -u # include href urls on links
67	agent-browser snapshot -i -c # compact (no empty structural nodes)
68	agent-browser snapshot -i -d 3 # cap depth at 3 levels
69	agent-browser snapshot -s "#main" # scope to a CSS selector
70	agent-browser snapshot -i --json # machine-readable output
71	```
72
73	Snapshot output looks like:
74
75	```
76	Page: Example - Log in
77	URL: https://example.com/login
78
79	@e1 [heading] "Log i

People who liked this also viewed

6 picks

Type