How do I install the Validate agent-browser for Web Automation skill?

Run npx skillmds add tools-only/validate-agent-browser-for-web-automation in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Validate agent-browser for Web Automation safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: makes network calls. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Validate agent-browser for Web Automation?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Validate agent-browser for Web Automation free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Validate agent-browser for Web Automation?

tools-only (@tools-only) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Validate agent-browser for Web Automation by tools-only

Name: Validate agent-browser for Web Automation
Rating: 4.5 (4 reviews)
Author: tools-only

Validate agent-browser for Web Automation

Test agent-browser (Playwright-based) on host with unrestricted network and Playwright browsers installed. **Validation steps**: - Install browsers: `npx playwright install` - Test: `npx agent-browser open https://code.claude.com/docs/en/skills` - Test: `npx agent-browser snapshot -i` (get element refs) - Test: `npx agent-browser get text body` (extract page text) - Verify snapshot/interact/re-snapshot workflow works - Document prerequisites for skill to function **Blocked on 2026-02-05**: Could not download Playwright browsers (DNS resolution failed, missing system libs) **Skill location**: `.claude/skills/agent-browser/SKILL.md`

tools-only 4635ff0 3 files · 8.5 KB Updated Sep 9, 2026 7 repo stars

File contents

Fact-Check

Claims checked: 5 VERIFIED: 4 | REFUTED: 1 | INCONCLUSIVE: 0

agent-browser is a real npm package — VERIFIED (v0.15.0, Vercel Labs, playwright-core dependency)
npx agent-browser open <url> works — VERIFIED (confirmed via --help)
npx agent-browser snapshot -i gets element refs — VERIFIED (documented with -i, --interactive flag)
npx agent-browser get text body extracts page text — REFUTED — documented syntax is get text <selector> where selector is @e1 ref or CSS selector. body works as CSS selector but is not the documented pattern.
snapshot/interact/re-snapshot workflow — VERIFIED (core workflow in SKILL.md and --help examples)

Refuted claims: get text body syntax — misleading but functional (body is a CSS selector) Citations: npm registry (v0.15.0), npx agent-browser --help output, GitHub repository docs (accessed 2026-02-26)

RT-ICA

Goal: Verify agent-browser skill works end-to-end on a properly equipped host and document prerequisites.

#	Condition	Status	Evidence
1	agent-browser CLI available	AVAILABLE	v0.15.0 via npx
2	Playwright browsers installed	AVAILABLE	Chromium 141.0 at /root/.cache/ms-playwright/
3	Network access for test URLs	DERIVABLE	Environment has network; target URL accessibility unverified
4	Skill documentation exists	AVAILABLE	SKILL.md + 7 references + 3 templates
5	Validation step syntax correct	MISSING	`get text body` refuted — should use `get text <selector>` with @ref or CSS
6	System libs for rendering	DERIVABLE	Headless mode should work; headed may need X11

Decision: APPROVED — no blockers for headless validation Missing: Update validation step to use get text @e1 or get text body (as CSS selector) with clarifying note

Groomed (2026-02-26)

Reproducibility

Install agent-browser: npm install -g agent-browser (or use npx)
Install Playwright browsers: agent-browser install (or npx playwright install chromium)
Open a test URL: agent-browser open https://example.com
Take interactive snapshot: agent-browser snapshot -i
Extract text via element ref: agent-browser get text @e1 (ref from snapshot output)
Chain open + wait + snapshot: agent-browser open https://example.com && agent-browser wait --load networkidle && agent-browser snapshot -i
Close browser: agent-browser close

Dependencies

Node.js 18+ with npx
Playwright Chromium browser binary (installed via agent-browser install)
Network access to target URLs
No external backlog items blocking this work

---name: validate-agent-browser-for-web-automationdescription: Test agent-browser (Playwright-based) on host with unrestricted network and Playwright browsers installed. **Validation steps**: - Install browsers: `npx playwright install` - Test: `npx agent-browser open https://code.claude.com/docs/en/skills` - Test: `npx agent-browser snapshot -i` (get element refs) - Test: `npx agent-browser get text body` (extract page text) - Verify snapshot/interact/re-snapshot workflow works - Document prerequisites for skill to function **Blocked on 2026-02-05**: Could not download Playwright browsers (DNS resolution failed, missing system libs) **Skill location**: `.claude/skills/agent-browser/SKILL.md`--- ## Fact-Check Claims checked: 5 VERIFIED: 4 | REFUTED: 1 | INCONCLUSIVE: 0 1. agent-browser is a real npm package — **VERIFIED** (v0.15.0, Vercel Labs, playwright-core dependency) 2. `npx agent-browser open <url>` works — **VERIFIED** (confirmed via `--help`) 3. `npx agent-browser snapshot -i` gets element refs — **VERIFIED** (documented with `-i, --interactive` flag) 4. `npx agent-browser get text body` extracts page text — **REFUTED** — documented syntax is `get text <selector>` where selector is `@e1` ref or CSS selector. `body` works as CSS selector but is not the documented pattern. 5. snapshot/interact/re-snapshot workflow — **VERIFIED** (core workflow in SKILL.md and `--help` examples) Refuted claims: `get text body` syntax — misleading but functional (body is a CSS selector) Citations: npm registry (v0.15.0), `npx agent-browser --help` output, GitHub repository docs (accessed 2026-02-26) ## RT-ICA **Goal**: Verify agent-browser skill works end-to-end on a properly equipped host and document prerequisites. | # | Condition | Status | Evidence | |---|-----------|--------|----------| | 1 | agent-browser CLI available | AVAILABLE | v0.15.0 via npx | | 2 | Playwright browsers installed | AVAILABLE | Chromium 141.0 at /root/.cache/ms-playwright/ | | 3 | Network access for test URLs | DERIVABLE | Environment has network; target URL accessibility unverified | | 4 | Skill documentation exists | AVAILABLE | SKILL.md + 7 references + 3 templates | | 5 | Validation step syntax correct | MISSING | `get text body` refuted — should use `get text <selector>` with @ref or CSS | | 6 | System libs for rendering | DERIVABLE | Headless mode should work; headed may need X11 | **Decision**: APPROVED — no blockers for headless validation **Missing**: Update validation step to use `get text @e1` or `get text body` (as CSS selector) with clarifying note ## Groomed (2026-02-26) ### Reproducibility 1. Install agent-browser: `npm install -g agent-browser` (or use `npx`) 2. Install Playwright browsers: `agent-browser install` (or `npx playwright install chromium`) 3. Open a test URL: `agent-browser open https://example.com` 4. Take interactive snapshot: `agent-browser snapshot -i` 5. Extract text via element ref: `agent-browser get text @e1` (ref from snapshot output) 6. Chain open + wait + snapshot: `agent-browser open https://example.com && agent-browser wait --load networkidle && agent-browser snapshot -i` 7. Close browser: `agent-browser close` ### Dependencies - Node.js 18+ with npx - Playwright Chromium browser binary (installed via `agent-browser install`) - Network access to target URLs - No external backlog items blocking this work

Frequently asked questions