A browser-control toolkit for AI agents that uses a real Chrome or Chromium browser. It can navigate pages, interact with forms, inspect sites, take screenshots, run checks, and extract structured data.

Use it to test localhost apps, fill in forms, audit websites, capture screenshots, collect page data, and compare visual changes.

It gives an agent a way to work with live websites and verify what appears in the browser. It also supports saved page references, human takeover, and evidence collection for browser tasks.

gsd-browser — Skill for Claude Code by open-gsd

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add open-gsd/gsd-browser --skill npm

Clone the repo

git clone --depth 1 https://github.com/open-gsd/gsd-browser

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/open-gsd/gsd-browser/npm/github.svg)](https://agentmods.dev/skills/open-gsd/gsd-browser/npm)

Your own site

<a href="https://agentmods.dev/skills/open-gsd/gsd-browser/npm"><img src="https://agentmods.dev/badge/skills/open-gsd/gsd-browser/npm/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/open-gsd/gsd-browser/npm"><img src="https://agentmods.dev/badge/skills/open-gsd/gsd-browser/npm.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 180 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 7,995 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin 94% copy Near-identical to another mod in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00180	$0.07995
Opus 5.5	$0.00072	$0.03198
Sonnet 5.5	$0.00036	$0.01599
Haiku 4.5	$0.00018	$0.00800

Measured 2mo ago against content hash 82d90cdfe593, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

gsd-browser scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.

The scan reads SKILL.md. This mod also ships 4 executable files (scripts/build-package.js, scripts/postinstall.js, scripts/test-package.js, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

This is a copy

94% identical to gsd-browser — 46 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

npm/SKILL.md · 842 lines

How it starts

The opening of the file, as written. The whole thing — 842 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Browser Automation with gsd-browser

Primary path for AI agents in 2026+: Use the MCP server — gsd-browser mcp.

It exposes 50+ tools, live resources (real snapshot/refs/state data), and executable prompts over stdio. All the unique strengths (versioned refs, live viewer + human collaboration/takeover/annotations/recordings, semantic intents, evidence bundles, self-healing action cache, batch execution, vault, visual regression, safety scanning, etc.) are available automatically to Cursor, Claude Desktop, VS Code + Copilot, and any other MCP client.

Quickstart for agents:

gsd-browser mcp

Then point your MCP client at it (see ./scripts/mcp-quickstart.sh cursor or claude/vscode for tailored configs).

Essential reading for MCP agents:

docs/mcp.md — Architecture, capabilities, client config examples, quickstart script.
docs/AGENT-BEST-PRACTICES.md — Golden rules, recommended workflows (login, audit, human-in-the-loop, evidence, self-healing), "When to Use What" table, response envelope usage, prompt/resource patterns.
docs/examples/mcp-client-config.json — Ready example.

The detailed command reference, error recovery, and workflow patterns in this SKILL.md document the exact semantics of the MCP tools (they are a direct 1:1 mapping). Use this as the source of truth when an MCP tool needs deeper clarification.

The installable curated skill pack lives in gsd-browser-skill/ (installed via the main installer or gsd-browser skill install).

Critical Rules (apply to both direct CLI and MCP usage)

The daemon auto-starts on browser commands. daemon health only reports state; it does not start a session. Use daemon start only when you want to pre-warm or verify daemon lifecycle explicitly.
Always re-snapshot after page changes. Refs are versioned (@v1:e1). After navigation, form submission, or dynamic content loading, old refs are stale. Run gsd-browser snapshot (or read the gsd-browser://latest-snapshot MCP resource) to get fresh refs.
Use --json when parsing output. Use text mode when reading output yourself. Use --json when you need to extract values programmatically (e.g., checking assertion results, parsing snapshot refs). MCP tools return rich envelopes containing structured_data.
Positional args have no flag prefix. Commands like click, type, hover take positional args — do NOT add --selector. See exact syntax in command reference below. (MCP equivalents: browser_click_ref etc. take ref as a simple string param.)
Use batch for atomic multi-step flows. Batch reduces round trips and keeps pass/fail checks in one call. Use separate commands when you need intermediate output (e.g., snapshot to discover refs). MCP equivalent: browser_batch.
Use view when the user wants to watch or direct the browser. The live viewer is an authenticated local workbench with Control, Annotate, Record, and Sensitive modes. Keep CLI/MCP commands on the same named session. MCP tools: browser_view, browser_takeover, browser_annotation_request, browser_record_*, browser_goal, etc.

Read the full file on GitHub · 842 lines

Files

What ships with it

11 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2mo ago First seen · 842 lines · 180 tokens per session scan A 82d90cdfe593

Subscribe to this mod's changes

gsd-browser is a skill published in the GitHub repository open-gsd/gsd-browser (57 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 180 tokens to every session and 7,995 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 94% identical to gsd-browser, differing in 46 lines, and is treated as a copy.

Browser automation

gsd-browser next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
gsd-browseropen-gsd/gsd-browser	Skill	57	1mo ago	Apache-2.0	A	180	Claude Code
interactive-loginyc-software/qm	Skill	16k	yesterday	MIT	A	46	Claude Code, Codex
pinchtab-mcppinchtab/pinchtab	Skill	11k	10d ago	MIT	A	52	Claude Code, Codex
azure-messaging-webpubsub-javamicrosoft/skills	Skill	3.2k	yesterday	MIT	A	43	Claude Code
cua-driverdavidondrej/skills	Skill	4.3k	3d ago	MIT	A	68	Claude Code, Codex
open-design-browsercompozy/compozy	Skill	2.8k	yesterday	MIT	A	56	Claude Code, Codex
southwestborski/travel-hacking-toolkit	Skill	895	yesterday	MIT	A	40	Claude Code