steel-mcp-server
MCP Serversteel-dev/steel-mcp-server
An MCP server that integrates Steel-managed Chromium browsers into AI clients for web scraping, screenshots, and interactive tasks.
49 16MITUpdated 2026-08-13
76
GreatOverview
The Steel MCP Server bridges AI assistants with headless and managed Chromium browsers via Steel's platform. It exposes stateless scraping tools along with advanced session-based interactive capabilities, allowing agents to navigate, click, type, and capture accessibility trees while supporting human-in-the-loop handoffs for sensitive tasks like logins and form fills.
Capabilities
- ▸Stateless web page scraping and markdown/accessibility tree extraction
- ▸Browser session creation, navigation, and management
- ▸Element finding via text, regex, or accessibility roles
- ▸Simulated user interactions including clicking, typing, and scrolling
- ▸Interactive live-view rendering and human-in-the-loop session handoffs
- ▸PDF rendering and screenshot capture
Best for
Scraping dynamic web pages and extracting pricing tables or articles, Navigating interactive web applications and filling out forms, Taking screenshots or rendering PDFs of target URLs, Performing authenticated actions securely through human-in-the-loop session handoffs
Works with
Claude CodeCursorGitHub Copilot
Evoa Score breakdown= Σ (score × weight)
76Task usefulness20%100 → +20.0
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%90 → +13.5
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%100 → +15.0
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%30 → +3.6
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%50 → +7.5
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%85 → +8.5
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%48 → +3.8
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%75 → +3.8
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total75.7 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
Credentials, passwords, and sensitive cookies are never exposed to the model context. Human-in-the-loop handoffs ensure users handle authentication walls directly.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.