agent-browser
Skill0 starsChanged 4 months agoMITView on GitHub
Browser automation CLI for AI agents. Use for website interaction, form automation, screenshots, scraping, and web app verification. Prefer snapshot refs (@e1, @e2) for deterministic actions.
- Installs packages
189 other projects use a version of this.
Adopted by
Most recently taken up by
- chrysogonus/power-agentstheir own version, first seen Nov 8, 2026
- kinhluan/skillstheir own version, first seen Nov 8, 2026
- cafe3310/public-agent-skillstheir own version, first seen Nov 8, 2026
- kortix-ai/sunatheir own version, first seen Nov 8, 2026
- HeteroCat/heterocat-skillstheir own version, first seen Nov 8, 2026
- JuanMarchetto/agent-skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use time-skill, dep-refs and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
Recepay6358Skilltime-skill
Display the current time in Pakistan Standard Time (PKT, UTC+5). Use when the user asks for the current time, Pakistan time, or PKT.
Also in 6 projects that use this
betacatslingSkilldep-refs
Use when needs to inspect TypeScript export references in the z-code workspace, list exports from a file, verify whether an export is unused before deletion, investigate who imports a symbol during refactors, or combine pnpm knip unused-export results with pnpm dep:refs symbol-level reference tracing.
Also in 5 projects that use this
betacatslingSkillfeature-boundary-planner
Map a ZCode behavior change to current UI surfaces, state owners, protocol commands, persistence, and validation. Use for impact analysis, product-boundary planning, or an implementation handoff grounded in the checked-out source.
Also in 5 projects that use this
nicepkgSkillcommunity-led-growth
Expert in community-led growth (CLG) - leveraging user communities to drive acquisition, retention, and expansion. Covers building developer communities, user groups, ambassador programs, and turning customers into advocates. Knows the difference between community as a feature vs community as a growth engine, and how to measure community ROI. Use when "community-led, CLG, ambassador program, champion program, user community, developer community, word of mouth, user group, " mentioned.
Also in 5 projects that use this
nicepkgSkillph-community-outreach
Engage Reddit, Indie Hackers, Hacker News and other communities for Product Hunt launches. Use this skill to identify relevant communities, craft appropriate posts, and drive cross-platform engagement without appearing spammy.
Also in 5 projects that use this
Recepay6358Skillweather-svg-creator
Creates an SVG weather card showing the current temperature for Dubai. Writes the SVG to orchestration-workflow/weather.svg and updates orchestration-workflow/output.md.
Also in 6 projects that use this
- +3 more from nicepkg
Other versions189 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 189
- shanraisshan/claude-code-best-practice★ 69ksame name
- moeru-ai/airi★ 51ksame name
- vercel-labs/agent-browser★ 47ksame name
- activepieces/activepieces★ 25ksame name
- slopus/happy★ 25ksame name
- kortix-ai/suna★ 21ksame name
- kortix-ai/CompanyOS★ 21ksame name
- eosphoros-ai/DB-GPT★ 20ksame name
- documenso/documenso★ 16ksame name
- emdash-cms/emdash★ 14ksame name
179 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- claude-real-videoSkill
- instalarSkill
- investSkill
- youtubeSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 173 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 68
Use encadeamento && quando a saída intermediária não é necessária.
- Distinctive: line 90
Use evidência explícita depois de cada ação.
- Distinctive: line 110
Use os controles de saída em páginas longas para não inflar o contexto.
“Common” is measured: kept by at least 6 in 10 of the 173 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
agent-browser - Browser Automation for AI Agents
Quando usar esta skill
- Abrir sites e automatizar ações de UI
- Preencher formulários, clicar em controles e verificar resultados
- Capturar screenshots/PDFs ou extrair conteúdo
- Rodar checagens web determinísticas usando refs de acessibilidade
- Executar tarefas de browser em paralelo com sessões isoladas
No Video AI Toolkit ela entra quando o print/captura exige interação antes
(fechar banner de cookies, logar, clicar numa aba, rolar até um gráfico, aguardar
conteúdo dinâmico) — casos que o web-shot.js de uma tacada não cobre. Para print
simples de uma URL, continue usando o web-shot.
Instalação
O CLI agent-browser é um binário à parte, instalado via npm (não vem vendorizado):
npm install -g agent-browserO npm 10+ pode bloquear o postinstall por segurança. Se agent-browser --version
falhar depois do install, rode o postinstall à mão (ele baixa o binário nativo do
release oficial e usa o Chrome/Edge já instalado — não baixa Chromium):
# a partir da pasta global do pacote
node scripts/postinstall.jsO install.ps1/install.sh do toolkit já fazem isso automaticamente.
Fluxo principal
Sempre use o loop determinístico de refs:
agent-browser open <url>agent-browser snapshot -i- interaja com refs (
@e1,@e2, ...) agent-browser snapshot -ide novo depois de mudanças de página/DOM
agent-browser open https://example.com/form
agent-browser wait --load networkidle
agent-browser snapshot -i
agent-browser fill @e1 "user@example.com"
agent-browser click @e2
agent-browser snapshot -iPadrões de comando
Use encadeamento && quando a saída intermediária não é necessária.
# Bom encadeamento: open -> wait -> snapshot
agent-browser open https://example.com && agent-browser wait --load networkidle && agent-browser snapshot -i
# Chamadas separadas quando a saída é necessária antes
agent-browser snapshot -i
# parse dos refs
agent-browser click @e2Comandos de alto valor:
- Navegação:
open,close - Snapshot:
snapshot -i,snapshot -i -C,snapshot -s "#selector" - Interação:
click,fill,type,select,check,press - Verificação:
diff snapshot,diff screenshot --baseline <file> - Captura:
screenshot,screenshot --annotate,pdf - Espera:
wait --load networkidle,wait <selector|@ref|ms>
Verificação
Use evidência explícita depois de cada ação.
# Baseline -> ação -> verifica estrutura
agent-browser snapshot -i
agent-browser click @e3
agent-browser diff snapshot
# Regressão visual
agent-browser screenshot baseline.png
agent-browser click @e5
agent-browser diff screenshot --baseline baseline.pngSegurança e confiabilidade
- Refs ficam inválidos após navegação ou mudança grande de DOM; re-snapshot antes da próxima ação.
- Prefira
wait --load networkidleou espera por selector/ref asleepfixo. - Para JS multi-passo, use
eval --stdin(ou base64) para evitar quebra de escape no shell. - Para tarefas concorrentes, isole com
--session <name>. - Use os controles de saída em páginas longas para não inflar o contexto.
- Endurecimento opcional em fluxos sensíveis: allowlist de domínio e políticas de ação.
Exemplos de endurecimento:
# Envolve o conteúdo da página em fronteiras (reduz risco de prompt-injection)
export AGENT_BROWSER_CONTENT_BOUNDARIES=1
# Limita o volume de saída em páginas longas
export AGENT_BROWSER_MAX_OUTPUT=50000
# Restringe navegação e rede a domínios confiáveis
export AGENT_BROWSER_ALLOWED_DOMAINS="example.com,*.example.com"
# Restringe os tipos de ação permitidos
export AGENT_BROWSER_ACTION_POLICY=./policy.jsonpolicy.json de exemplo:
{"default":"deny","allow":["navigate","snapshot","click","fill","scroll","wait","get"],"deny":["eval","download","upload","network","state"]}Equivalente por flag de CLI:
agent-browser --content-boundaries --max-output 50000 --allowed-domains "example.com,*.example.com" --action-policy ./policy.json open https://example.comNotas de plataforma (Windows / antivírus)
Duas pegadinhas conhecidas ao rodar num terminal automatizado no Windows:
opené um daemon.agent-browser open <url>sobe o browser e fica vivo de propósito para os comandos seguintes reusarem a sessão — ele não "retorna". Num terminal automatizado (não-interativo), chamaropende forma síncrona parece travar, mas não travou: dispare oopendestacado (em background), espere o processo do browser subir e rodesnapshot/click/closecomo chamadas separadas curtas. Num terminal humano normal isso não é problema.readx antivírus que intercepta HTTPS. O comandoagent-browser read <url>faz uma requisição HTTP nativa (fora do browser). Se um antivírus estiver interceptando HTTPS (Avast, Kaspersky etc.), ela falha cominvalid peer certificate: UnknownIssuer. O modo browser (open+snapshot) usa o Chrome do sistema, que confia no certificado do antivírus, e não sofre disso — prefira esse caminho quando oreadfalhar.
Troubleshooting
command not found: instale comnpm install -g agent-browsere rode o postinstall.- Elemento errado clicado: rode
snapshot -ide novo e use refs frescos. - Conteúdo de SPA dinâmico faltando: espere com
--load networkidleouwaitpor selector. - Colisão de sessão: use nomes
--sessionúnicos e feche cada sessão. - Pressão de saída grande: estreite snapshots (
-i,-c,-d,-s) e extraia só o necessário.
Referências
Docs aprofundados nesta skill:
- commands
- snapshot-refs
- session-management
- authentication
Recursos relacionados:
Templates prontos:
./templates/form-automation.sh./templates/capture-workflow.sh
Metadata
- Version: 1.1.0
- Scope: automação de browser determinística para fluxos de agente
- Origem:
vercel-labs/agent-browser(MIT) — verATTRIBUTION.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

